/[pcre]/code/trunk/pcre_jit_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_jit_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1249 by zherczeg, Mon Feb 18 09:55:43 2013 UTC revision 1452 by zherczeg, Tue Jan 28 16:07:52 2014 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2013 University of Cambridge
10    
11    The machine code generator part (this module) was written by Zoltan Herczeg    The machine code generator part (this module) was written by Zoltan Herczeg
12                        Copyright (c) 2010-2012                        Copyright (c) 2010-2013
13    
14  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
15  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 71  system files. */ Line 71  system files. */
71     2 - Enable capture_last_ptr (includes option 1). */     2 - Enable capture_last_ptr (includes option 1). */
72  /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */  /* #define DEBUG_FORCE_UNOPTIMIZED_CBRAS 2 */
73    
74    /* 1 - Always have a control head. */
75    /* #define DEBUG_FORCE_CONTROL_HEAD 1 */
76    
77  /* Allocate memory for the regex stack on the real machine stack.  /* Allocate memory for the regex stack on the real machine stack.
78  Fast, but limited size. */  Fast, but limited size. */
79  #define MACHINE_STACK_SIZE 32768  #define MACHINE_STACK_SIZE 32768
# Line 165  typedef struct jit_arguments { Line 168  typedef struct jit_arguments {
168    pcre_uchar *mark_ptr;    pcre_uchar *mark_ptr;
169    void *callout_data;    void *callout_data;
170    /* Everything else after. */    /* Everything else after. */
171      pcre_uint32 limit_match;
172      int real_offset_count;
173    int offset_count;    int offset_count;
   int call_limit;  
174    pcre_uint8 notbol;    pcre_uint8 notbol;
175    pcre_uint8 noteol;    pcre_uint8 noteol;
176    pcre_uint8 notempty;    pcre_uint8 notempty;
# Line 175  typedef struct jit_arguments { Line 179  typedef struct jit_arguments {
179    
180  typedef struct executable_functions {  typedef struct executable_functions {
181    void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES];    void *executable_funcs[JIT_NUMBER_OF_COMPILE_MODES];
182      sljit_uw *read_only_data[JIT_NUMBER_OF_COMPILE_MODES];
183      sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];
184    PUBL(jit_callback) callback;    PUBL(jit_callback) callback;
185    void *userdata;    void *userdata;
186    pcre_uint32 top_bracket;    pcre_uint32 top_bracket;
187    sljit_uw executable_sizes[JIT_NUMBER_OF_COMPILE_MODES];    pcre_uint32 limit_match;
188  } executable_functions;  } executable_functions;
189    
190  typedef struct jump_list {  typedef struct jump_list {
# Line 192  typedef struct stub_list { Line 198  typedef struct stub_list {
198    struct stub_list *next;    struct stub_list *next;
199  } stub_list;  } stub_list;
200    
201  enum frame_types { no_frame = -1, no_stack = -2 };  typedef struct label_addr_list {
202      struct sljit_label *label;
203      sljit_uw *addr;
204      struct label_addr_list *next;
205    } label_addr_list;
206    
207    enum frame_types {
208      no_frame = -1,
209      no_stack = -2
210    };
211    
212    enum control_types {
213      type_mark = 0,
214      type_then_trap = 1
215    };
216    
217  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);  typedef int (SLJIT_CALL *jit_function)(jit_arguments *args);
218    
219  /* The following structure is the key data type for the recursive  /* The following structure is the key data type for the recursive
220  code generator. It is allocated by compile_matchingpath, and contains  code generator. It is allocated by compile_matchingpath, and contains
221  the aguments for compile_backtrackingpath. Must be the first member  the arguments for compile_backtrackingpath. Must be the first member
222  of its descendants. */  of its descendants. */
223  typedef struct backtrack_common {  typedef struct backtrack_common {
224    /* Concatenation stack. */    /* Concatenation stack. */
# Line 214  typedef struct backtrack_common { Line 234  typedef struct backtrack_common {
234  typedef struct assert_backtrack {  typedef struct assert_backtrack {
235    backtrack_common common;    backtrack_common common;
236    jump_list *condfailed;    jump_list *condfailed;
237    /* Less than 0 (-1) if a frame is not needed. */    /* Less than 0 if a frame is not needed. */
238    int framesize;    int framesize;
239    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
240    int private_data_ptr;    int private_data_ptr;
# Line 235  typedef struct bracket_backtrack { Line 255  typedef struct bracket_backtrack {
255      /* Both for OP_COND, OP_SCOND. */      /* Both for OP_COND, OP_SCOND. */
256      jump_list *condfailed;      jump_list *condfailed;
257      assert_backtrack *assert;      assert_backtrack *assert;
258      /* For OP_ONCE. -1 if not needed. */      /* For OP_ONCE. Less than 0 if not needed. */
259      int framesize;      int framesize;
260    } u;    } u;
261    /* Points to our private memory word on the stack. */    /* Points to our private memory word on the stack. */
# Line 270  typedef struct recurse_entry { Line 290  typedef struct recurse_entry {
290    /* Collects the calls until the function is not created. */    /* Collects the calls until the function is not created. */
291    jump_list *calls;    jump_list *calls;
292    /* Points to the starting opcode. */    /* Points to the starting opcode. */
293    int start;    sljit_sw start;
294  } recurse_entry;  } recurse_entry;
295    
296  typedef struct recurse_backtrack {  typedef struct recurse_backtrack {
# Line 278  typedef struct recurse_backtrack { Line 298  typedef struct recurse_backtrack {
298    BOOL inlined_pattern;    BOOL inlined_pattern;
299  } recurse_backtrack;  } recurse_backtrack;
300    
301  #define MAX_RANGE_SIZE 6  #define OP_THEN_TRAP OP_TABLE_LENGTH
302    
303    typedef struct then_trap_backtrack {
304      backtrack_common common;
305      /* If then_trap is not NULL, this structure contains the real
306      then_trap for the backtracking path. */
307      struct then_trap_backtrack *then_trap;
308      /* Points to the starting opcode. */
309      sljit_sw start;
310      /* Exit point for the then opcodes of this alternative. */
311      jump_list *quit;
312      /* Frame size of the current alternative. */
313      int framesize;
314    } then_trap_backtrack;
315    
316    #define MAX_RANGE_SIZE 4
317    
318  typedef struct compiler_common {  typedef struct compiler_common {
319      /* The sljit ceneric compiler. */
320    struct sljit_compiler *compiler;    struct sljit_compiler *compiler;
321      /* First byte code. */
322    pcre_uchar *start;    pcre_uchar *start;
   
323    /* Maps private data offset to each opcode. */    /* Maps private data offset to each opcode. */
324    int *private_data_ptrs;    sljit_si *private_data_ptrs;
325      /* This read-only data is available during runtime. */
326      sljit_uw *read_only_data;
327      /* The total size of the read-only data. */
328      sljit_uw read_only_data_size;
329      /* The next free entry of the read_only_data. */
330      sljit_uw *read_only_data_ptr;
331    /* Tells whether the capturing bracket is optimized. */    /* Tells whether the capturing bracket is optimized. */
332    pcre_uint8 *optimized_cbracket;    pcre_uint8 *optimized_cbracket;
333      /* Tells whether the starting offset is a target of then. */
334      pcre_uint8 *then_offsets;
335      /* Current position where a THEN must jump. */
336      then_trap_backtrack *then_trap;
337    /* Starting offset of private data for capturing brackets. */    /* Starting offset of private data for capturing brackets. */
338    int cbraptr;    int cbra_ptr;
339    /* OVector starting point. Must be divisible by 2. */    /* Output vector starting point. Must be divisible by 2. */
340    int ovector_start;    int ovector_start;
341    /* Last known position of the requested byte. */    /* Last known position of the requested byte. */
342    int req_char_ptr;    int req_char_ptr;
# Line 304  typedef struct compiler_common { Line 350  typedef struct compiler_common {
350    int first_line_end;    int first_line_end;
351    /* Points to the marked string. */    /* Points to the marked string. */
352    int mark_ptr;    int mark_ptr;
353      /* Recursive control verb management chain. */
354      int control_head_ptr;
355    /* Points to the last matched capture block index. */    /* Points to the last matched capture block index. */
356    int capture_last_ptr;    int capture_last_ptr;
357      /* Points to the starting position of the current match. */
358      int start_ptr;
359    
360    /* Flipped and lower case tables. */    /* Flipped and lower case tables. */
361    const pcre_uint8 *fcc;    const pcre_uint8 *fcc;
362    sljit_sw lcc;    sljit_sw lcc;
363    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */    /* Mode can be PCRE_STUDY_JIT_COMPILE and others. */
364    int mode;    int mode;
365      /* TRUE, when minlength is greater than 0. */
366      BOOL might_be_empty;
367      /* \K is found in the pattern. */
368      BOOL has_set_som;
369      /* (*SKIP:arg) is found in the pattern. */
370      BOOL has_skip_arg;
371      /* (*THEN) is found in the pattern. */
372      BOOL has_then;
373      /* Needs to know the start position anytime. */
374      BOOL needs_start_ptr;
375      /* Currently in recurse or negative assert. */
376      BOOL local_exit;
377      /* Currently in a positive assert. */
378      BOOL positive_assert;
379    /* Newline control. */    /* Newline control. */
380    int nltype;    int nltype;
381      pcre_uint32 nlmax;
382      pcre_uint32 nlmin;
383    int newline;    int newline;
384    int bsr_nltype;    int bsr_nltype;
385      pcre_uint32 bsr_nlmax;
386      pcre_uint32 bsr_nlmin;
387    /* Dollar endonly. */    /* Dollar endonly. */
388    int endonly;    int endonly;
   BOOL has_set_som;  
389    /* Tables. */    /* Tables. */
390    sljit_sw ctypes;    sljit_sw ctypes;
   int digits[2 + MAX_RANGE_SIZE];  
391    /* Named capturing brackets. */    /* Named capturing brackets. */
392    sljit_uw name_table;    pcre_uchar *name_table;
393    sljit_sw name_count;    sljit_sw name_count;
394    sljit_sw name_entry_size;    sljit_sw name_entry_size;
395    
# Line 332  typedef struct compiler_common { Line 398  typedef struct compiler_common {
398    struct sljit_label *quit_label;    struct sljit_label *quit_label;
399    struct sljit_label *forced_quit_label;    struct sljit_label *forced_quit_label;
400    struct sljit_label *accept_label;    struct sljit_label *accept_label;
401      struct sljit_label *ff_newline_shortcut;
402    stub_list *stubs;    stub_list *stubs;
403      label_addr_list *label_addrs;
404    recurse_entry *entries;    recurse_entry *entries;
405    recurse_entry *currententry;    recurse_entry *currententry;
406    jump_list *partialmatch;    jump_list *partialmatch;
407    jump_list *quit;    jump_list *quit;
408      jump_list *positive_assert_quit;
409    jump_list *forced_quit;    jump_list *forced_quit;
410    jump_list *accept;    jump_list *accept;
411    jump_list *calllimit;    jump_list *calllimit;
# Line 348  typedef struct compiler_common { Line 417  typedef struct compiler_common {
417    jump_list *vspace;    jump_list *vspace;
418    jump_list *casefulcmp;    jump_list *casefulcmp;
419    jump_list *caselesscmp;    jump_list *caselesscmp;
420      jump_list *reset_match;
421    BOOL jscript_compat;    BOOL jscript_compat;
422  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
423    BOOL utf;    BOOL utf;
424  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
425    BOOL use_ucp;    BOOL use_ucp;
426  #endif  #endif
 #ifndef COMPILE_PCRE32  
   jump_list *utfreadchar;  
 #endif  
427  #ifdef COMPILE_PCRE8  #ifdef COMPILE_PCRE8
428      jump_list *utfreadchar;
429      jump_list *utfreadchar16;
430    jump_list *utfreadtype8;    jump_list *utfreadtype8;
431  #endif  #endif
432  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
# Line 414  typedef struct compare_context { Line 483  typedef struct compare_context {
483  #define STACK_TOP     SLJIT_SCRATCH_REG2  #define STACK_TOP     SLJIT_SCRATCH_REG2
484  #define STACK_LIMIT   SLJIT_SAVED_REG3  #define STACK_LIMIT   SLJIT_SAVED_REG3
485  #define ARGUMENTS     SLJIT_SAVED_EREG1  #define ARGUMENTS     SLJIT_SAVED_EREG1
486  #define CALL_COUNT    SLJIT_SAVED_EREG2  #define COUNT_MATCH   SLJIT_SAVED_EREG2
487  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1  #define RETURN_ADDR   SLJIT_TEMPORARY_EREG1
488    
489  /* Local space layout. */  /* Local space layout. */
# Line 425  typedef struct compare_context { Line 494  typedef struct compare_context {
494  #define POSSESSIVE0      (2 * sizeof(sljit_sw))  #define POSSESSIVE0      (2 * sizeof(sljit_sw))
495  #define POSSESSIVE1      (3 * sizeof(sljit_sw))  #define POSSESSIVE1      (3 * sizeof(sljit_sw))
496  /* Max limit of recursions. */  /* Max limit of recursions. */
497  #define CALL_LIMIT       (4 * sizeof(sljit_sw))  #define LIMIT_MATCH      (4 * sizeof(sljit_sw))
498  /* The output vector is stored on the stack, and contains pointers  /* The output vector is stored on the stack, and contains pointers
499  to characters. The vector data is divided into two groups: the first  to characters. The vector data is divided into two groups: the first
500  group contains the start / end character pointers, and the second is  group contains the start / end character pointers, and the second is
501  the start pointers when the end of the capturing group has not yet reached. */  the start pointers when the end of the capturing group has not yet reached. */
502  #define OVECTOR_START    (common->ovector_start)  #define OVECTOR_START    (common->ovector_start)
503  #define OVECTOR(i)       (OVECTOR_START + (i) * sizeof(sljit_sw))  #define OVECTOR(i)       (OVECTOR_START + (i) * (sljit_sw)sizeof(sljit_sw))
504  #define OVECTOR_PRIV(i)  (common->cbraptr + (i) * sizeof(sljit_sw))  #define OVECTOR_PRIV(i)  (common->cbra_ptr + (i) * (sljit_sw)sizeof(sljit_sw))
505  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])  #define PRIVATE_DATA(cc) (common->private_data_ptrs[(cc) - common->start])
506    
507  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
# Line 474  the start pointers when the end of the c Line 543  the start pointers when the end of the c
543  #define GET_LOCAL_BASE(dst, dstw, offset) \  #define GET_LOCAL_BASE(dst, dstw, offset) \
544    sljit_get_local_base(compiler, (dst), (dstw), (offset))    sljit_get_local_base(compiler, (dst), (dstw), (offset))
545    
546    #define READ_CHAR_MAX 0x7fffffff
547    
548  static pcre_uchar* bracketend(pcre_uchar* cc)  static pcre_uchar* bracketend(pcre_uchar* cc)
549  {  {
550  SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND));  SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND));
# Line 483  cc += 1 + LINK_SIZE; Line 554  cc += 1 + LINK_SIZE;
554  return cc;  return cc;
555  }  }
556    
557    static int no_alternatives(pcre_uchar* cc)
558    {
559    int count = 0;
560    SLJIT_ASSERT((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND));
561    do
562      {
563      cc += GET(cc, 1);
564      count++;
565      }
566    while (*cc == OP_ALT);
567    SLJIT_ASSERT(*cc >= OP_KET && *cc <= OP_KETRPOS);
568    return count;
569    }
570    
571    static int ones_in_half_byte[16] = {
572      /* 0 */ 0, 1, 1, 2, /* 4 */ 1, 2, 2, 3,
573      /* 8 */ 1, 2, 2, 3, /* 12 */ 2, 3, 3, 4
574    };
575    
576  /* Functions whose might need modification for all new supported opcodes:  /* Functions whose might need modification for all new supported opcodes:
577   next_opcode   next_opcode
578   get_private_data_length   check_opcode_types
579   set_private_data_ptrs   set_private_data_ptrs
580   get_framesize   get_framesize
581   init_frame   init_frame
582   get_private_data_length_for_copy   get_private_data_copy_length
583   copy_private_data   copy_private_data
584   compile_matchingpath   compile_matchingpath
585   compile_backtrackingpath   compile_backtrackingpath
# Line 535  switch(*cc) Line 625  switch(*cc)
625    case OP_CRMINQUERY:    case OP_CRMINQUERY:
626    case OP_CRRANGE:    case OP_CRRANGE:
627    case OP_CRMINRANGE:    case OP_CRMINRANGE:
628      case OP_CRPOSSTAR:
629      case OP_CRPOSPLUS:
630      case OP_CRPOSQUERY:
631      case OP_CRPOSRANGE:
632    case OP_CLASS:    case OP_CLASS:
633    case OP_NCLASS:    case OP_NCLASS:
634    case OP_REF:    case OP_REF:
635    case OP_REFI:    case OP_REFI:
636      case OP_DNREF:
637      case OP_DNREFI:
638    case OP_RECURSE:    case OP_RECURSE:
639    case OP_CALLOUT:    case OP_CALLOUT:
640    case OP_ALT:    case OP_ALT:
# Line 564  switch(*cc) Line 660  switch(*cc)
660    case OP_SCBRAPOS:    case OP_SCBRAPOS:
661    case OP_SCOND:    case OP_SCOND:
662    case OP_CREF:    case OP_CREF:
663    case OP_NCREF:    case OP_DNCREF:
664    case OP_RREF:    case OP_RREF:
665    case OP_NRREF:    case OP_DNRREF:
666    case OP_DEF:    case OP_DEF:
667    case OP_BRAZERO:    case OP_BRAZERO:
668    case OP_BRAMINZERO:    case OP_BRAMINZERO:
669    case OP_BRAPOSZERO:    case OP_BRAPOSZERO:
670      case OP_PRUNE:
671      case OP_SKIP:
672      case OP_THEN:
673    case OP_COMMIT:    case OP_COMMIT:
674    case OP_FAIL:    case OP_FAIL:
675    case OP_ACCEPT:    case OP_ACCEPT:
# Line 669  switch(*cc) Line 768  switch(*cc)
768  #endif  #endif
769    
770    case OP_MARK:    case OP_MARK:
771      case OP_PRUNE_ARG:
772      case OP_SKIP_ARG:
773      case OP_THEN_ARG:
774    return cc + 1 + 2 + cc[1];    return cc + 1 + 2 + cc[1];
775    
776    default:    default:
777      /* All opcodes are supported now! */
778      SLJIT_ASSERT_STOP();
779    return NULL;    return NULL;
780    }    }
781  }  }
782    
783    static BOOL check_opcode_types(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)
784    {
785    int count;
786    pcre_uchar *slot;
787    
788    /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */
789    while (cc < ccend)
790      {
791      switch(*cc)
792        {
793        case OP_SET_SOM:
794        common->has_set_som = TRUE;
795        common->might_be_empty = TRUE;
796        cc += 1;
797        break;
798    
799        case OP_REF:
800        case OP_REFI:
801        common->optimized_cbracket[GET2(cc, 1)] = 0;
802        cc += 1 + IMM2_SIZE;
803        break;
804    
805        case OP_BRA:
806        case OP_CBRA:
807        case OP_SBRA:
808        case OP_SCBRA:
809        count = no_alternatives(cc);
810        if (count > 4)
811          common->read_only_data_size += count * sizeof(sljit_uw);
812        cc += 1 + LINK_SIZE + (*cc == OP_CBRA || *cc == OP_SCBRA ? IMM2_SIZE : 0);
813        break;
814    
815        case OP_CBRAPOS:
816        case OP_SCBRAPOS:
817        common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;
818        cc += 1 + LINK_SIZE + IMM2_SIZE;
819        break;
820    
821        case OP_COND:
822        case OP_SCOND:
823        /* Only AUTO_CALLOUT can insert this opcode. We do
824           not intend to support this case. */
825        if (cc[1 + LINK_SIZE] == OP_CALLOUT)
826          return FALSE;
827        cc += 1 + LINK_SIZE;
828        break;
829    
830        case OP_CREF:
831        common->optimized_cbracket[GET2(cc, 1)] = 0;
832        cc += 1 + IMM2_SIZE;
833        break;
834    
835        case OP_DNREF:
836        case OP_DNREFI:
837        case OP_DNCREF:
838        count = GET2(cc, 1 + IMM2_SIZE);
839        slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
840        while (count-- > 0)
841          {
842          common->optimized_cbracket[GET2(slot, 0)] = 0;
843          slot += common->name_entry_size;
844          }
845        cc += 1 + 2 * IMM2_SIZE;
846        break;
847    
848        case OP_RECURSE:
849        /* Set its value only once. */
850        if (common->recursive_head_ptr == 0)
851          {
852          common->recursive_head_ptr = common->ovector_start;
853          common->ovector_start += sizeof(sljit_sw);
854          }
855        cc += 1 + LINK_SIZE;
856        break;
857    
858        case OP_CALLOUT:
859        if (common->capture_last_ptr == 0)
860          {
861          common->capture_last_ptr = common->ovector_start;
862          common->ovector_start += sizeof(sljit_sw);
863          }
864        cc += 2 + 2 * LINK_SIZE;
865        break;
866    
867        case OP_THEN_ARG:
868        common->has_then = TRUE;
869        common->control_head_ptr = 1;
870        /* Fall through. */
871    
872        case OP_PRUNE_ARG:
873        common->needs_start_ptr = TRUE;
874        /* Fall through. */
875    
876        case OP_MARK:
877        if (common->mark_ptr == 0)
878          {
879          common->mark_ptr = common->ovector_start;
880          common->ovector_start += sizeof(sljit_sw);
881          }
882        cc += 1 + 2 + cc[1];
883        break;
884    
885        case OP_THEN:
886        common->has_then = TRUE;
887        common->control_head_ptr = 1;
888        /* Fall through. */
889    
890        case OP_PRUNE:
891        case OP_SKIP:
892        common->needs_start_ptr = TRUE;
893        cc += 1;
894        break;
895    
896        case OP_SKIP_ARG:
897        common->control_head_ptr = 1;
898        common->has_skip_arg = TRUE;
899        cc += 1 + 2 + cc[1];
900        break;
901    
902        default:
903        cc = next_opcode(common, cc);
904        if (cc == NULL)
905          return FALSE;
906        break;
907        }
908      }
909    return TRUE;
910    }
911    
912    static int get_class_iterator_size(pcre_uchar *cc)
913    {
914    switch(*cc)
915      {
916      case OP_CRSTAR:
917      case OP_CRPLUS:
918      return 2;
919    
920      case OP_CRMINSTAR:
921      case OP_CRMINPLUS:
922      case OP_CRQUERY:
923      case OP_CRMINQUERY:
924      return 1;
925    
926      case OP_CRRANGE:
927      case OP_CRMINRANGE:
928      if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))
929        return 0;
930      return 2;
931    
932      default:
933      return 0;
934      }
935    }
936    
937    static BOOL detect_repeat(compiler_common *common, pcre_uchar *begin)
938    {
939    pcre_uchar *end = bracketend(begin);
940    pcre_uchar *next;
941    pcre_uchar *next_end;
942    pcre_uchar *max_end;
943    pcre_uchar type;
944    sljit_sw length = end - begin;
945    int min, max, i;
946    
947    /* Detect fixed iterations first. */
948    if (end[-(1 + LINK_SIZE)] != OP_KET)
949      return FALSE;
950    
951    /* Already detected repeat. */
952    if (common->private_data_ptrs[end - common->start - LINK_SIZE] != 0)
953      return TRUE;
954    
955    next = end;
956    min = 1;
957    while (1)
958      {
959      if (*next != *begin)
960        break;
961      next_end = bracketend(next);
962      if (next_end - next != length || memcmp(begin, next, IN_UCHARS(length)) != 0)
963        break;
964      next = next_end;
965      min++;
966      }
967    
968    if (min == 2)
969      return FALSE;
970    
971    max = 0;
972    max_end = next;
973    if (*next == OP_BRAZERO || *next == OP_BRAMINZERO)
974      {
975      type = *next;
976      while (1)
977        {
978        if (next[0] != type || next[1] != OP_BRA || next[2 + LINK_SIZE] != *begin)
979          break;
980        next_end = bracketend(next + 2 + LINK_SIZE);
981        if (next_end - next != (length + 2 + LINK_SIZE) || memcmp(begin, next + 2 + LINK_SIZE, IN_UCHARS(length)) != 0)
982          break;
983        next = next_end;
984        max++;
985        }
986    
987      if (next[0] == type && next[1] == *begin && max >= 1)
988        {
989        next_end = bracketend(next + 1);
990        if (next_end - next == (length + 1) && memcmp(begin, next + 1, IN_UCHARS(length)) == 0)
991          {
992          for (i = 0; i < max; i++, next_end += 1 + LINK_SIZE)
993            if (*next_end != OP_KET)
994              break;
995    
996          if (i == max)
997            {
998            common->private_data_ptrs[max_end - common->start - LINK_SIZE] = next_end - max_end;
999            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 1] = (type == OP_BRAZERO) ? OP_UPTO : OP_MINUPTO;
1000            /* +2 the original and the last. */
1001            common->private_data_ptrs[max_end - common->start - LINK_SIZE + 2] = max + 2;
1002            if (min == 1)
1003              return TRUE;
1004            min--;
1005            max_end -= (1 + LINK_SIZE) + GET(max_end, -LINK_SIZE);
1006            }
1007          }
1008        }
1009      }
1010    
1011    if (min >= 3)
1012      {
1013      common->private_data_ptrs[end - common->start - LINK_SIZE] = max_end - end;
1014      common->private_data_ptrs[end - common->start - LINK_SIZE + 1] = OP_EXACT;
1015      common->private_data_ptrs[end - common->start - LINK_SIZE + 2] = min;
1016      return TRUE;
1017      }
1018    
1019    return FALSE;
1020    }
1021    
1022  #define CASE_ITERATOR_PRIVATE_DATA_1 \  #define CASE_ITERATOR_PRIVATE_DATA_1 \
1023      case OP_MINSTAR: \      case OP_MINSTAR: \
1024      case OP_MINPLUS: \      case OP_MINPLUS: \
# Line 728  switch(*cc) Line 1071  switch(*cc)
1071      case OP_TYPEUPTO: \      case OP_TYPEUPTO: \
1072      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
1073    
1074  static int get_class_iterator_size(pcre_uchar *cc)  static void set_private_data_ptrs(compiler_common *common, int *private_data_start, pcre_uchar *ccend)
 {  
 switch(*cc)  
   {  
   case OP_CRSTAR:  
   case OP_CRPLUS:  
   return 2;  
   
   case OP_CRMINSTAR:  
   case OP_CRMINPLUS:  
   case OP_CRQUERY:  
   case OP_CRMINQUERY:  
   return 1;  
   
   case OP_CRRANGE:  
   case OP_CRMINRANGE:  
   if (GET2(cc, 1) == GET2(cc, 1 + IMM2_SIZE))  
     return 0;  
   return 2;  
   
   default:  
   return 0;  
   }  
 }  
   
 static int get_private_data_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)  
1075  {  {
1076  int private_data_length = 0;  pcre_uchar *cc = common->start;
1077  pcre_uchar *alternative;  pcre_uchar *alternative;
 pcre_uchar *name;  
1078  pcre_uchar *end = NULL;  pcre_uchar *end = NULL;
1079  int space, size, i;  int private_data_ptr = *private_data_start;
1080  pcre_uint32 bracketlen;  int space, size, bracketlen;
1081    
 /* Calculate important variables (like stack size) and checks whether all opcodes are supported. */  
1082  while (cc < ccend)  while (cc < ccend)
1083    {    {
1084    space = 0;    space = 0;
1085    size = 0;    size = 0;
1086    bracketlen = 0;    bracketlen = 0;
1087      if (private_data_ptr > SLJIT_MAX_LOCAL_SIZE)
1088        return;
1089    
1090      if (*cc == OP_ONCE || *cc == OP_ONCE_NC || *cc == OP_BRA || *cc == OP_CBRA || *cc == OP_COND)
1091        if (detect_repeat(common, cc))
1092          {
1093          /* These brackets are converted to repeats, so no global
1094          based single character repeat is allowed. */
1095          if (cc >= end)
1096            end = bracketend(cc);
1097          }
1098    
1099    switch(*cc)    switch(*cc)
1100      {      {
1101      case OP_SET_SOM:      case OP_KET:
1102      common->has_set_som = TRUE;      if (common->private_data_ptrs[cc + 1 - common->start] != 0)
1103      cc += 1;        {
1104      break;        common->private_data_ptrs[cc - common->start] = private_data_ptr;
1105          private_data_ptr += sizeof(sljit_sw);
1106      case OP_REF:        cc += common->private_data_ptrs[cc + 1 - common->start];
1107      case OP_REFI:        }
1108      common->optimized_cbracket[GET2(cc, 1)] = 0;      cc += 1 + LINK_SIZE;
     cc += 1 + IMM2_SIZE;  
1109      break;      break;
1110    
1111      case OP_ASSERT:      case OP_ASSERT:
# Line 790  while (cc < ccend) Line 1117  while (cc < ccend)
1117      case OP_BRAPOS:      case OP_BRAPOS:
1118      case OP_SBRA:      case OP_SBRA:
1119      case OP_SBRAPOS:      case OP_SBRAPOS:
1120      private_data_length += sizeof(sljit_sw);      case OP_SCOND:
1121        common->private_data_ptrs[cc - common->start] = private_data_ptr;
1122        private_data_ptr += sizeof(sljit_sw);
1123      bracketlen = 1 + LINK_SIZE;      bracketlen = 1 + LINK_SIZE;
1124      break;      break;
1125    
1126      case OP_CBRAPOS:      case OP_CBRAPOS:
1127      case OP_SCBRAPOS:      case OP_SCBRAPOS:
1128      private_data_length += sizeof(sljit_sw);      common->private_data_ptrs[cc - common->start] = private_data_ptr;
1129      common->optimized_cbracket[GET2(cc, 1 + LINK_SIZE)] = 0;      private_data_ptr += sizeof(sljit_sw);
1130      bracketlen = 1 + LINK_SIZE + IMM2_SIZE;      bracketlen = 1 + LINK_SIZE + IMM2_SIZE;
1131      break;      break;
1132    
1133      case OP_COND:      case OP_COND:
1134      case OP_SCOND:      /* Might be a hidden SCOND. */
1135      /* Only AUTO_CALLOUT can insert this opcode. We do      alternative = cc + GET(cc, 1);
1136         not intend to support this case. */      if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)
     if (cc[1 + LINK_SIZE] == OP_CALLOUT)  
       return -1;  
   
     if (*cc == OP_COND)  
1137        {        {
1138        /* Might be a hidden SCOND. */        common->private_data_ptrs[cc - common->start] = private_data_ptr;
1139        alternative = cc + GET(cc, 1);        private_data_ptr += sizeof(sljit_sw);
       if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)  
         private_data_length += sizeof(sljit_sw);  
       }  
     else  
       private_data_length += sizeof(sljit_sw);  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
     case OP_CREF:  
     i = GET2(cc, 1);  
     common->optimized_cbracket[i] = 0;  
     cc += 1 + IMM2_SIZE;  
     break;  
   
     case OP_NCREF:  
     bracketlen = GET2(cc, 1);  
     name = (pcre_uchar *)common->name_table;  
     alternative = name;  
     for (i = 0; i < common->name_count; i++)  
       {  
       if (GET2(name, 0) == bracketlen) break;  
       name += common->name_entry_size;  
       }  
     SLJIT_ASSERT(i != common->name_count);  
   
     for (i = 0; i < common->name_count; i++)  
       {  
       if (STRCMP_UC_UC(alternative + IMM2_SIZE, name + IMM2_SIZE) == 0)  
         common->optimized_cbracket[GET2(alternative, 0)] = 0;  
       alternative += common->name_entry_size;  
       }  
     bracketlen = 0;  
     cc += 1 + IMM2_SIZE;  
     break;  
   
     case OP_BRA:  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
     case OP_CBRA:  
     case OP_SCBRA:  
     bracketlen = 1 + LINK_SIZE + IMM2_SIZE;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_1  
     space = 1;  
     size = -2;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_2A  
     space = 2;  
     size = -2;  
     break;  
   
     CASE_ITERATOR_PRIVATE_DATA_2B  
     space = 2;  
     size = -(2 + IMM2_SIZE);  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_1  
     space = 1;  
     size = 1;  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_2A  
     if (cc[1] != OP_ANYNL && cc[1] != OP_EXTUNI)  
       space = 2;  
     size = 1;  
     break;  
   
     CASE_ITERATOR_TYPE_PRIVATE_DATA_2B  
     if (cc[1 + IMM2_SIZE] != OP_ANYNL && cc[1 + IMM2_SIZE] != OP_EXTUNI)  
       space = 2;  
     size = 1 + IMM2_SIZE;  
     break;  
   
     case OP_CLASS:  
     case OP_NCLASS:  
     size += 1 + 32 / sizeof(pcre_uchar);  
     space = get_class_iterator_size(cc + size);  
     break;  
   
 #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  
     case OP_XCLASS:  
     size = GET(cc, 1);  
     space = get_class_iterator_size(cc + size);  
     break;  
 #endif  
   
     case OP_RECURSE:  
     /* Set its value only once. */  
     if (common->recursive_head_ptr == 0)  
       {  
       common->recursive_head_ptr = common->ovector_start;  
       common->ovector_start += sizeof(sljit_sw);  
       }  
     cc += 1 + LINK_SIZE;  
     break;  
   
     case OP_CALLOUT:  
     if (common->capture_last_ptr == 0)  
       {  
       common->capture_last_ptr = common->ovector_start;  
       common->ovector_start += sizeof(sljit_sw);  
       }  
     cc += 2 + 2 * LINK_SIZE;  
     break;  
   
     case OP_MARK:  
     if (common->mark_ptr == 0)  
       {  
       common->mark_ptr = common->ovector_start;  
       common->ovector_start += sizeof(sljit_sw);  
       }  
     cc += 1 + 2 + cc[1];  
     break;  
   
     default:  
     cc = next_opcode(common, cc);  
     if (cc == NULL)  
       return -1;  
     break;  
     }  
   
   if (space > 0 && cc >= end)  
     private_data_length += sizeof(sljit_sw) * space;  
   
   if (size != 0)  
     {  
     if (size < 0)  
       {  
       cc += -size;  
 #ifdef SUPPORT_UTF  
       if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
       }  
     else  
       cc += size;  
     }  
   
   if (bracketlen != 0)  
     {  
     if (cc >= end)  
       {  
       end = bracketend(cc);  
       if (end[-1 - LINK_SIZE] == OP_KET)  
         end = NULL;  
       }  
     cc += bracketlen;  
     }  
   }  
 return private_data_length;  
 }  
   
 static void set_private_data_ptrs(compiler_common *common, int private_data_ptr, pcre_uchar *ccend)  
 {  
 pcre_uchar *cc = common->start;  
 pcre_uchar *alternative;  
 pcre_uchar *end = NULL;  
 int space, size, bracketlen;  
   
 while (cc < ccend)  
   {  
   space = 0;  
   size = 0;  
   bracketlen = 0;  
   switch(*cc)  
     {  
     case OP_ASSERT:  
     case OP_ASSERT_NOT:  
     case OP_ASSERTBACK:  
     case OP_ASSERTBACK_NOT:  
     case OP_ONCE:  
     case OP_ONCE_NC:  
     case OP_BRAPOS:  
     case OP_SBRA:  
     case OP_SBRAPOS:  
     case OP_SCOND:  
     common->private_data_ptrs[cc - common->start] = private_data_ptr;  
     private_data_ptr += sizeof(sljit_sw);  
     bracketlen = 1 + LINK_SIZE;  
     break;  
   
     case OP_CBRAPOS:  
     case OP_SCBRAPOS:  
     common->private_data_ptrs[cc - common->start] = private_data_ptr;  
     private_data_ptr += sizeof(sljit_sw);  
     bracketlen = 1 + LINK_SIZE + IMM2_SIZE;  
     break;  
   
     case OP_COND:  
     /* Might be a hidden SCOND. */  
     alternative = cc + GET(cc, 1);  
     if (*alternative == OP_KETRMAX || *alternative == OP_KETRMIN)  
       {  
       common->private_data_ptrs[cc - common->start] = private_data_ptr;  
       private_data_ptr += sizeof(sljit_sw);  
1140        }        }
1141      bracketlen = 1 + LINK_SIZE;      bracketlen = 1 + LINK_SIZE;
1142      break;      break;
# Line 1073  while (cc < ccend) Line 1201  while (cc < ccend)
1201      break;      break;
1202      }      }
1203    
1204      /* Character iterators, which are not inside a repeated bracket,
1205         gets a private slot instead of allocating it on the stack. */
1206    if (space > 0 && cc >= end)    if (space > 0 && cc >= end)
1207      {      {
1208      common->private_data_ptrs[cc - common->start] = private_data_ptr;      common->private_data_ptrs[cc - common->start] = private_data_ptr;
# Line 1103  while (cc < ccend) Line 1233  while (cc < ccend)
1233      cc += bracketlen;      cc += bracketlen;
1234      }      }
1235    }    }
1236    *private_data_start = private_data_ptr;
1237  }  }
1238    
1239  /* Returns with a frame_types (always < 0) if no need for frame. */  /* Returns with a frame_types (always < 0) if no need for frame. */
1240  static int get_framesize(compiler_common *common, pcre_uchar *cc, BOOL recursive)  static int get_framesize(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL recursive, BOOL* needs_control_head)
1241  {  {
 pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE);  
1242  int length = 0;  int length = 0;
1243  int possessive = 0;  int possessive = 0;
1244  BOOL stack_restore = FALSE;  BOOL stack_restore = FALSE;
# Line 1117  BOOL setmark_found = recursive; Line 1247  BOOL setmark_found = recursive;
1247  /* The last capture is a local variable even for recursions. */  /* The last capture is a local variable even for recursions. */
1248  BOOL capture_last_found = FALSE;  BOOL capture_last_found = FALSE;
1249    
1250  if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))  #if defined DEBUG_FORCE_CONTROL_HEAD && DEBUG_FORCE_CONTROL_HEAD
1251    SLJIT_ASSERT(common->control_head_ptr != 0);
1252    *needs_control_head = TRUE;
1253    #else
1254    *needs_control_head = FALSE;
1255    #endif
1256    
1257    if (ccend == NULL)
1258    {    {
1259    possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;    ccend = bracketend(cc) - (1 + LINK_SIZE);
1260    /* This is correct regardless of common->capture_last_ptr. */    if (!recursive && (*cc == OP_CBRAPOS || *cc == OP_SCBRAPOS))
1261    capture_last_found = TRUE;      {
1262        possessive = length = (common->capture_last_ptr != 0) ? 5 : 3;
1263        /* This is correct regardless of common->capture_last_ptr. */
1264        capture_last_found = TRUE;
1265        }
1266      cc = next_opcode(common, cc);
1267    }    }
1268    
 cc = next_opcode(common, cc);  
1269  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1270  while (cc < ccend)  while (cc < ccend)
1271    switch(*cc)    switch(*cc)
# Line 1141  while (cc < ccend) Line 1282  while (cc < ccend)
1282      break;      break;
1283    
1284      case OP_MARK:      case OP_MARK:
1285        case OP_PRUNE_ARG:
1286        case OP_THEN_ARG:
1287      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1288      stack_restore = TRUE;      stack_restore = TRUE;
1289      if (!setmark_found)      if (!setmark_found)
# Line 1148  while (cc < ccend) Line 1291  while (cc < ccend)
1291        length += 2;        length += 2;
1292        setmark_found = TRUE;        setmark_found = TRUE;
1293        }        }
1294        if (common->control_head_ptr != 0)
1295          *needs_control_head = TRUE;
1296      cc += 1 + 2 + cc[1];      cc += 1 + 2 + cc[1];
1297      break;      break;
1298    
# Line 1267  if (length > 0) Line 1412  if (length > 0)
1412  return stack_restore ? no_frame : no_stack;  return stack_restore ? no_frame : no_stack;
1413  }  }
1414    
1415  static void init_frame(compiler_common *common, pcre_uchar *cc, int stackpos, int stacktop, BOOL recursive)  static void init_frame(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, int stackpos, int stacktop, BOOL recursive)
1416  {  {
1417  DEFINE_COMPILER;  DEFINE_COMPILER;
 pcre_uchar *ccend = bracketend(cc) - (1 + LINK_SIZE);  
1418  BOOL setsom_found = recursive;  BOOL setsom_found = recursive;
1419  BOOL setmark_found = recursive;  BOOL setmark_found = recursive;
1420  /* The last capture is a local variable even for recursions. */  /* The last capture is a local variable even for recursions. */
# Line 1282  SLJIT_UNUSED_ARG(stacktop); Line 1426  SLJIT_UNUSED_ARG(stacktop);
1426  SLJIT_ASSERT(stackpos >= stacktop + 2);  SLJIT_ASSERT(stackpos >= stacktop + 2);
1427    
1428  stackpos = STACK(stackpos);  stackpos = STACK(stackpos);
1429  if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))  if (ccend == NULL)
1430    cc = next_opcode(common, cc);    {
1431      ccend = bracketend(cc) - (1 + LINK_SIZE);
1432      if (recursive || (*cc != OP_CBRAPOS && *cc != OP_SCBRAPOS))
1433        cc = next_opcode(common, cc);
1434      }
1435    
1436  SLJIT_ASSERT(cc != NULL);  SLJIT_ASSERT(cc != NULL);
1437  while (cc < ccend)  while (cc < ccend)
1438    switch(*cc)    switch(*cc)
# Line 1303  while (cc < ccend) Line 1452  while (cc < ccend)
1452      break;      break;
1453    
1454      case OP_MARK:      case OP_MARK:
1455        case OP_PRUNE_ARG:
1456        case OP_THEN_ARG:
1457      SLJIT_ASSERT(common->mark_ptr != 0);      SLJIT_ASSERT(common->mark_ptr != 0);
1458      if (!setmark_found)      if (!setmark_found)
1459        {        {
# Line 1383  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st Line 1534  OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), st
1534  SLJIT_ASSERT(stackpos == STACK(stacktop));  SLJIT_ASSERT(stackpos == STACK(stacktop));
1535  }  }
1536    
1537  static SLJIT_INLINE int get_private_data_length_for_copy(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend)  static SLJIT_INLINE int get_private_data_copy_length(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend, BOOL needs_control_head)
1538  {  {
1539  int private_data_length = 2;  int private_data_length = needs_control_head ? 3 : 2;
1540  int size;  int size;
1541  pcre_uchar *alternative;  pcre_uchar *alternative;
1542  /* Calculate the sum of the private machine words. */  /* Calculate the sum of the private machine words. */
# Line 1394  while (cc < ccend) Line 1545  while (cc < ccend)
1545    size = 0;    size = 0;
1546    switch(*cc)    switch(*cc)
1547      {      {
1548        case OP_KET:
1549        if (PRIVATE_DATA(cc) != 0)
1550          private_data_length++;
1551        cc += 1 + LINK_SIZE;
1552        break;
1553    
1554      case OP_ASSERT:      case OP_ASSERT:
1555      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1556      case OP_ASSERTBACK:      case OP_ASSERTBACK:
# Line 1498  return private_data_length; Line 1655  return private_data_length;
1655  }  }
1656    
1657  static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend,  static void copy_private_data(compiler_common *common, pcre_uchar *cc, pcre_uchar *ccend,
1658    BOOL save, int stackptr, int stacktop)    BOOL save, int stackptr, int stacktop, BOOL needs_control_head)
1659  {  {
1660  DEFINE_COMPILER;  DEFINE_COMPILER;
1661  int srcw[2];  int srcw[2];
# Line 1519  stacktop = STACK(stacktop - 1); Line 1676  stacktop = STACK(stacktop - 1);
1676    
1677  if (!save)  if (!save)
1678    {    {
1679    stackptr += sizeof(sljit_sw);    stackptr += (needs_control_head ? 2 : 1) * sizeof(sljit_sw);
1680    if (stackptr < stacktop)    if (stackptr < stacktop)
1681      {      {
1682      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(STACK_TOP), stackptr);
# Line 1535  if (!save) Line 1692  if (!save)
1692    /* The tmp1next must be TRUE in either way. */    /* The tmp1next must be TRUE in either way. */
1693    }    }
1694    
1695  while (status != end)  do
1696    {    {
1697    count = 0;    count = 0;
1698    switch(status)    switch(status)
# Line 1544  while (status != end) Line 1701  while (status != end)
1701      SLJIT_ASSERT(save && common->recursive_head_ptr != 0);      SLJIT_ASSERT(save && common->recursive_head_ptr != 0);
1702      count = 1;      count = 1;
1703      srcw[0] = common->recursive_head_ptr;      srcw[0] = common->recursive_head_ptr;
1704        if (needs_control_head)
1705          {
1706          SLJIT_ASSERT(common->control_head_ptr != 0);
1707          count = 2;
1708          srcw[1] = common->control_head_ptr;
1709          }
1710      status = loop;      status = loop;
1711      break;      break;
1712    
# Line 1556  while (status != end) Line 1719  while (status != end)
1719    
1720      switch(*cc)      switch(*cc)
1721        {        {
1722          case OP_KET:
1723          if (PRIVATE_DATA(cc) != 0)
1724            {
1725            count = 1;
1726            srcw[0] = PRIVATE_DATA(cc);
1727            }
1728          cc += 1 + LINK_SIZE;
1729          break;
1730    
1731        case OP_ASSERT:        case OP_ASSERT:
1732        case OP_ASSERT_NOT:        case OP_ASSERT_NOT:
1733        case OP_ASSERTBACK:        case OP_ASSERTBACK:
# Line 1768  while (status != end) Line 1940  while (status != end)
1940        }        }
1941      }      }
1942    }    }
1943    while (status != end);
1944    
1945  if (save)  if (save)
1946    {    {
# Line 1801  if (save) Line 1974  if (save)
1974  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));  SLJIT_ASSERT(cc == ccend && stackptr == stacktop && (save || (tmp1empty && tmp2empty)));
1975  }  }
1976    
1977    static SLJIT_INLINE pcre_uchar *set_then_offsets(compiler_common *common, pcre_uchar *cc, pcre_uint8 *current_offset)
1978    {
1979    pcre_uchar *end = bracketend(cc);
1980    BOOL has_alternatives = cc[GET(cc, 1)] == OP_ALT;
1981    
1982    /* Assert captures then. */
1983    if (*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT)
1984      current_offset = NULL;
1985    /* Conditional block does not. */
1986    if (*cc == OP_COND || *cc == OP_SCOND)
1987      has_alternatives = FALSE;
1988    
1989    cc = next_opcode(common, cc);
1990    if (has_alternatives)
1991      current_offset = common->then_offsets + (cc - common->start);
1992    
1993    while (cc < end)
1994      {
1995      if ((*cc >= OP_ASSERT && *cc <= OP_ASSERTBACK_NOT) || (*cc >= OP_ONCE && *cc <= OP_SCOND))
1996        cc = set_then_offsets(common, cc, current_offset);
1997      else
1998        {
1999        if (*cc == OP_ALT && has_alternatives)
2000          current_offset = common->then_offsets + (cc + 1 + LINK_SIZE - common->start);
2001        if (*cc >= OP_THEN && *cc <= OP_THEN_ARG && current_offset != NULL)
2002          *current_offset = 1;
2003        cc = next_opcode(common, cc);
2004        }
2005      }
2006    
2007    return end;
2008    }
2009    
2010  #undef CASE_ITERATOR_PRIVATE_DATA_1  #undef CASE_ITERATOR_PRIVATE_DATA_1
2011  #undef CASE_ITERATOR_PRIVATE_DATA_2A  #undef CASE_ITERATOR_PRIVATE_DATA_2A
2012  #undef CASE_ITERATOR_PRIVATE_DATA_2B  #undef CASE_ITERATOR_PRIVATE_DATA_2B
# Line 1864  while (list_item) Line 2070  while (list_item)
2070  common->stubs = NULL;  common->stubs = NULL;
2071  }  }
2072    
2073  static SLJIT_INLINE void decrease_call_count(compiler_common *common)  static void add_label_addr(compiler_common *common)
2074  {  {
2075  DEFINE_COMPILER;  DEFINE_COMPILER;
2076    label_addr_list *label_addr;
2077    
2078  OP2(SLJIT_SUB | SLJIT_SET_E, CALL_COUNT, 0, CALL_COUNT, 0, SLJIT_IMM, 1);  label_addr = sljit_alloc_memory(compiler, sizeof(label_addr_list));
2079    if (label_addr == NULL)
2080      return;
2081    label_addr->label = LABEL();
2082    label_addr->addr = common->read_only_data_ptr;
2083    label_addr->next = common->label_addrs;
2084    common->label_addrs = label_addr;
2085    common->read_only_data_ptr++;
2086    }
2087    
2088    static SLJIT_INLINE void count_match(compiler_common *common)
2089    {
2090    DEFINE_COMPILER;
2091    
2092    OP2(SLJIT_SUB | SLJIT_SET_E, COUNT_MATCH, 0, COUNT_MATCH, 0, SLJIT_IMM, 1);
2093  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));  add_jump(compiler, &common->calllimit, JUMP(SLJIT_C_ZERO));
2094  }  }
2095    
# Line 1899  static SLJIT_INLINE void reset_ovector(c Line 2120  static SLJIT_INLINE void reset_ovector(c
2120  DEFINE_COMPILER;  DEFINE_COMPILER;
2121  struct sljit_label *loop;  struct sljit_label *loop;
2122  int i;  int i;
2123    
2124  /* At this point we can freely use all temporary registers. */  /* At this point we can freely use all temporary registers. */
2125    SLJIT_ASSERT(length > 1);
2126  /* TMP1 returns with begin - 1. */  /* TMP1 returns with begin - 1. */
2127  OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_SUB, SLJIT_SCRATCH_REG1, 0, SLJIT_MEM1(SLJIT_SAVED_REG1), SLJIT_OFFSETOF(jit_arguments, begin), SLJIT_IMM, IN_UCHARS(1));
2128  if (length < 8)  if (length < 8)
2129    {    {
2130    for (i = 0; i < length; i++)    for (i = 1; i < length; i++)
2131      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), SLJIT_SCRATCH_REG1, 0);
2132    }    }
2133  else  else
2134    {    {
2135    GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START - sizeof(sljit_sw));    GET_LOCAL_BASE(SLJIT_SCRATCH_REG2, 0, OVECTOR_START);
2136    OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length);    OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, length - 1);
2137    loop = LABEL();    loop = LABEL();
2138    OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0);    OP1(SLJIT_MOVU, SLJIT_MEM1(SLJIT_SCRATCH_REG2), sizeof(sljit_sw), SLJIT_SCRATCH_REG1, 0);
2139    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1);    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 1);
# Line 1918  else Line 2141  else
2141    }    }
2142  }  }
2143    
2144    static SLJIT_INLINE void do_reset_match(compiler_common *common, int length)
2145    {
2146    DEFINE_COMPILER;
2147    struct sljit_label *loop;
2148    int i;
2149    
2150    SLJIT_ASSERT(length > 1);
2151    /* OVECTOR(1) contains the "string begin - 1" constant. */
2152    if (length > 2)
2153      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
2154    if (length < 8)
2155      {
2156      for (i = 2; i < length; i++)
2157        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(i), TMP1, 0);
2158      }
2159    else
2160      {
2161      GET_LOCAL_BASE(TMP2, 0, OVECTOR_START + sizeof(sljit_sw));
2162      OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_IMM, length - 2);
2163      loop = LABEL();
2164      OP1(SLJIT_MOVU, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
2165      OP2(SLJIT_SUB | SLJIT_SET_E, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, 1);
2166      JUMPTO(SLJIT_C_NOT_ZERO, loop);
2167      }
2168    
2169    OP1(SLJIT_MOV, STACK_TOP, 0, ARGUMENTS, 0);
2170    if (common->mark_ptr != 0)
2171      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mark_ptr, SLJIT_IMM, 0);
2172    if (common->control_head_ptr != 0)
2173      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->control_head_ptr, SLJIT_IMM, 0);
2174    OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(jit_arguments, stack));
2175    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_ptr);
2176    OP1(SLJIT_MOV, STACK_TOP, 0, SLJIT_MEM1(STACK_TOP), SLJIT_OFFSETOF(struct sljit_stack, base));
2177    }
2178    
2179    static sljit_sw SLJIT_CALL do_search_mark(sljit_sw *current, const pcre_uchar *skip_arg)
2180    {
2181    while (current != NULL)
2182      {
2183      switch (current[-2])
2184        {
2185        case type_then_trap:
2186        break;
2187    
2188        case type_mark:
2189        if (STRCMP_UC_UC(skip_arg, (pcre_uchar *)current[-3]) == 0)
2190          return current[-4];
2191        break;
2192    
2193        default:
2194        SLJIT_ASSERT_STOP();
2195        break;
2196        }
2197      current = (sljit_sw*)current[-1];
2198      }
2199    return -1;
2200    }
2201    
2202  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)  static SLJIT_INLINE void copy_ovector(compiler_common *common, int topbracket)
2203  {  {
2204  DEFINE_COMPILER;  DEFINE_COMPILER;
# Line 1971  else Line 2252  else
2252  static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit)  static SLJIT_INLINE void return_with_partial_match(compiler_common *common, struct sljit_label *quit)
2253  {  {
2254  DEFINE_COMPILER;  DEFINE_COMPILER;
2255    struct sljit_jump *jump;
2256    
2257  SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2);  SLJIT_COMPILE_ASSERT(STR_END == SLJIT_SAVED_REG2, str_end_must_be_saved_reg2);
2258  SLJIT_ASSERT(common->start_used_ptr != 0 && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0));  SLJIT_ASSERT(common->start_used_ptr != 0 && common->start_ptr != 0
2259      && (common->mode == JIT_PARTIAL_SOFT_COMPILE ? common->hit_start != 0 : common->hit_start == 0));
2260    
2261  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0);  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, ARGUMENTS, 0);
2262  OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL);  OP1(SLJIT_MOV, SLJIT_RETURN_REG, 0, SLJIT_IMM, PCRE_ERROR_PARTIAL);
2263  OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offset_count));  OP1(SLJIT_MOV_SI, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, real_offset_count));
2264  CMPTO(SLJIT_C_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit);  CMPTO(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 2, quit);
2265    
2266  /* Store match begin and end. */  /* Store match begin and end. */
2267  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin));  OP1(SLJIT_MOV, SLJIT_SAVED_REG1, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, begin));
2268  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets));  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG2, 0, SLJIT_MEM1(SLJIT_SCRATCH_REG2), SLJIT_OFFSETOF(jit_arguments, offsets));
2269    
2270    jump = CMP(SLJIT_C_SIG_LESS, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, 3);
2271    OP2(SLJIT_SUB, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_ptr : (common->hit_start + (int)sizeof(sljit_sw)), SLJIT_SAVED_REG1, 0);
2272    #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2273    OP2(SLJIT_ASHR, SLJIT_SCRATCH_REG3, 0, SLJIT_SCRATCH_REG3, 0, SLJIT_IMM, UCHAR_SHIFT);
2274    #endif
2275    OP1(SLJIT_MOV_SI, SLJIT_MEM1(SLJIT_SCRATCH_REG2), 2 * sizeof(int), SLJIT_SCRATCH_REG3, 0);
2276    JUMPHERE(jump);
2277    
2278  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start);  OP1(SLJIT_MOV, SLJIT_SCRATCH_REG3, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->mode == JIT_PARTIAL_HARD_COMPILE ? common->start_used_ptr : common->hit_start);
2279  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0);  OP2(SLJIT_SUB, SLJIT_SAVED_REG2, 0, STR_END, 0, SLJIT_SAVED_REG1, 0);
2280  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
# Line 2144  return (bit < 256) ? ((0 << 8) | bit) : Line 2436  return (bit < 256) ? ((0 << 8) | bit) :
2436    
2437  static void check_partial(compiler_common *common, BOOL force)  static void check_partial(compiler_common *common, BOOL force)
2438  {  {
2439  /* Checks whether a partial matching is occured. Does not modify registers. */  /* Checks whether a partial matching is occurred. Does not modify registers. */
2440  DEFINE_COMPILER;  DEFINE_COMPILER;
2441  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
2442    
# Line 2159  else if (common->mode == JIT_PARTIAL_SOF Line 2451  else if (common->mode == JIT_PARTIAL_SOF
2451    jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1);    jump = CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, SLJIT_IMM, -1);
2452    
2453  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2454    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2455  else  else
2456    {    {
2457    if (common->partialmatchlabel != NULL)    if (common->partialmatchlabel != NULL)
# Line 2172  if (jump != NULL) Line 2464  if (jump != NULL)
2464    JUMPHERE(jump);    JUMPHERE(jump);
2465  }  }
2466    
2467  static struct sljit_jump *check_str_end(compiler_common *common)  static void check_str_end(compiler_common *common, jump_list **end_reached)
2468  {  {
2469  /* Does not affect registers. Usually used in a tight spot. */  /* Does not affect registers. Usually used in a tight spot. */
2470  DEFINE_COMPILER;  DEFINE_COMPILER;
2471  struct sljit_jump *jump;  struct sljit_jump *jump;
 struct sljit_jump *nohit;  
 struct sljit_jump *return_value;  
2472    
2473  if (common->mode == JIT_COMPILE)  if (common->mode == JIT_COMPILE)
2474    return CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);    {
2475      add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
2476      return;
2477      }
2478    
2479  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0);  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0);
2480  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2481    {    {
2482    nohit = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0);    add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2483    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2484    JUMPHERE(nohit);    add_jump(compiler, end_reached, JUMP(SLJIT_JUMP));
   return_value = JUMP(SLJIT_JUMP);  
2485    }    }
2486  else  else
2487    {    {
2488    return_value = CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0);    add_jump(compiler, end_reached, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2489    if (common->partialmatchlabel != NULL)    if (common->partialmatchlabel != NULL)
2490      JUMPTO(SLJIT_JUMP, common->partialmatchlabel);      JUMPTO(SLJIT_JUMP, common->partialmatchlabel);
2491    else    else
2492      add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));      add_jump(compiler, &common->partialmatch, JUMP(SLJIT_JUMP));
2493    }    }
2494  JUMPHERE(jump);  JUMPHERE(jump);
 return return_value;  
2495  }  }
2496    
2497  static void detect_partial_match(compiler_common *common, jump_list **backtracks)  static void detect_partial_match(compiler_common *common, jump_list **backtracks)
# Line 2219  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR Line 2510  jump = CMP(SLJIT_C_LESS, STR_PTR, 0, STR
2510  add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));  add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), common->start_used_ptr, STR_PTR, 0));
2511  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)  if (common->mode == JIT_PARTIAL_SOFT_COMPILE)
2512    {    {
2513    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, -1);    OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->hit_start, SLJIT_IMM, 0);
2514    add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));    add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
2515    }    }
2516  else  else
# Line 2232  else Line 2523  else
2523  JUMPHERE(jump);  JUMPHERE(jump);
2524  }  }
2525    
2526  static void read_char(compiler_common *common)  static void peek_char(compiler_common *common, pcre_uint32 max)
2527  {  {
2528  /* Reads the character into TMP1, updates STR_PTR.  /* Reads the character into TMP1, keeps STR_PTR.
2529  Does not check STR_END. TMP2 Destroyed. */  Does not check STR_END. TMP2 Destroyed. */
2530  DEFINE_COMPILER;  DEFINE_COMPILER;
2531  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2532  struct sljit_jump *jump;  struct sljit_jump *jump;
2533  #endif  #endif
2534    
2535    SLJIT_UNUSED_ARG(max);
2536    
2537  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
2538  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2539  if (common->utf)  if (common->utf)
2540    {    {
2541  #if defined COMPILE_PCRE8    if (max < 128) return;
2542    
2543    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2544  #elif defined COMPILE_PCRE16    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
   jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);  
 #endif /* COMPILE_PCRE[8|16] */  
2545    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2546      OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2547    JUMPHERE(jump);    JUMPHERE(jump);
2548    }    }
2549  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */
2550    
2551    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2552    if (common->utf)
2553      {
2554      if (max < 0xd800) return;
2555    
2556      OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2557      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2558      /* TMP2 contains the high surrogate. */
2559      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2560      OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40);
2561      OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
2562      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);
2563      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2564      JUMPHERE(jump);
2565      }
2566    #endif
2567    }
2568    
2569    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2570    
2571    static BOOL is_char7_bitset(const pcre_uint8 *bitset, BOOL nclass)
2572    {
2573    /* Tells whether the character codes below 128 are enough
2574    to determine a match. */
2575    const pcre_uint8 value = nclass ? 0xff : 0;
2576    const pcre_uint8* end = bitset + 32;
2577    
2578    bitset += 16;
2579    do
2580      {
2581      if (*bitset++ != value)
2582        return FALSE;
2583      }
2584    while (bitset < end);
2585    return TRUE;
2586    }
2587    
2588    static void read_char7_type(compiler_common *common, BOOL full_read)
2589    {
2590    /* Reads the precise character type of a character into TMP1, if the character
2591    is less than 128. Otherwise it returns with zero. Does not check STR_END. The
2592    full_read argument tells whether characters above max are accepted or not. */
2593    DEFINE_COMPILER;
2594    struct sljit_jump *jump;
2595    
2596    SLJIT_ASSERT(common->utf);
2597    
2598    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2599  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2600    
2601    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2602    
2603    if (full_read)
2604      {
2605      jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);
2606      OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2607      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2608      JUMPHERE(jump);
2609      }
2610  }  }
2611    
2612  static void peek_char(compiler_common *common)  #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
2613    
2614    static void read_char_range(compiler_common *common, pcre_uint32 min, pcre_uint32 max, BOOL update_str_ptr)
2615  {  {
2616  /* Reads the character into TMP1, keeps STR_PTR.  /* Reads the precise value of a character into TMP1, if the character is
2617  Does not check STR_END. TMP2 Destroyed. */  between min and max (c >= min && c <= max). Otherwise it returns with a value
2618    outside the range. Does not check STR_END. */
2619  DEFINE_COMPILER;  DEFINE_COMPILER;
2620  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2621  struct sljit_jump *jump;  struct sljit_jump *jump;
2622  #endif  #endif
2623    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2624    struct sljit_jump *jump2;
2625    #endif
2626    
2627  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);  SLJIT_UNUSED_ARG(update_str_ptr);
2628  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32  SLJIT_UNUSED_ARG(min);
2629    SLJIT_UNUSED_ARG(max);
2630    SLJIT_ASSERT(min <= max);
2631    
2632    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2633    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2634    
2635    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2636  if (common->utf)  if (common->utf)
2637    {    {
2638  #if defined COMPILE_PCRE8    if (max < 128 && !update_str_ptr) return;
2639    
2640    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xc0);
2641  #elif defined COMPILE_PCRE16    if (min >= 0x10000)
2642    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xd800);      {
2643  #endif /* COMPILE_PCRE[8|16] */      OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xf0);
2644    add_jump(compiler, &common->utfreadchar, JUMP(SLJIT_FAST_CALL));      if (update_str_ptr)
2645    OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, TMP2, 0);        OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2646        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2647        jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0x7);
2648        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2649        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2650        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2651        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2652        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2653        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2654        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2655        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
2656        if (!update_str_ptr)
2657          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
2658        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2659        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2660        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2661        JUMPHERE(jump2);
2662        if (update_str_ptr)
2663          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
2664        }
2665      else if (min >= 0x800 && max <= 0xffff)
2666        {
2667        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xe0);
2668        if (update_str_ptr)
2669          OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2670        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2671        jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xf);
2672        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2673        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2674        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2675        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2676        if (!update_str_ptr)
2677          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
2678        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2679        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2680        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2681        JUMPHERE(jump2);
2682        if (update_str_ptr)
2683          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
2684        }
2685      else if (max >= 0x800)
2686        add_jump(compiler, (max < 0x10000) ? &common->utfreadchar16 : &common->utfreadchar, JUMP(SLJIT_FAST_CALL));
2687      else if (max < 128)
2688        {
2689        OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2690        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2691        }
2692      else
2693        {
2694        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2695        if (!update_str_ptr)
2696          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2697        else
2698          OP1(SLJIT_MOV_UB, RETURN_ADDR, 0, SLJIT_MEM1(TMP1), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2699        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2700        OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2701        OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2702        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2703        if (update_str_ptr)
2704          OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, RETURN_ADDR, 0);
2705        }
2706    JUMPHERE(jump);    JUMPHERE(jump);
2707    }    }
2708  #endif /* SUPPORT_UTF && !COMPILE_PCRE32 */  #endif
2709    
2710    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2711    if (common->utf)
2712      {
2713      if (max >= 0x10000)
2714        {
2715        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2716        jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2717        /* TMP2 contains the high surrogate. */
2718        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2719        OP2(SLJIT_ADD, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x40);
2720        OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 10);
2721        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2722        OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);
2723        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2724        JUMPHERE(jump);
2725        return;
2726        }
2727    
2728      if (max < 0xd800 && !update_str_ptr) return;
2729    
2730      /* Skip low surrogate if necessary. */
2731      OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, 0xd800);
2732      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2733      if (update_str_ptr)
2734        OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2735      if (max >= 0xd800)
2736        OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0x10000);
2737      JUMPHERE(jump);
2738      }
2739    #endif
2740  }  }
2741    
2742  static void read_char8_type(compiler_common *common)  static SLJIT_INLINE void read_char(compiler_common *common)
2743    {
2744    read_char_range(common, 0, READ_CHAR_MAX, TRUE);
2745    }
2746    
2747    static void read_char8_type(compiler_common *common, BOOL update_str_ptr)
2748  {  {
2749  /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */  /* Reads the character type into TMP1, updates STR_PTR. Does not check STR_END. */
2750  DEFINE_COMPILER;  DEFINE_COMPILER;
2751  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2752  struct sljit_jump *jump;  struct sljit_jump *jump;
2753  #endif  #endif
2754    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2755    struct sljit_jump *jump2;
2756    #endif
2757    
2758  #ifdef SUPPORT_UTF  SLJIT_UNUSED_ARG(update_str_ptr);
2759    
2760    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);
2761    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2762    
2763    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
2764  if (common->utf)  if (common->utf)
2765    {    {
   OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);  
   OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
 #if defined COMPILE_PCRE8  
2766    /* This can be an extra read in some situations, but hopefully    /* This can be an extra read in some situations, but hopefully
2767    it is needed in most cases. */    it is needed in most cases. */
2768    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2769    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);    jump = CMP(SLJIT_C_LESS, TMP2, 0, SLJIT_IMM, 0xc0);
2770    add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));    if (!update_str_ptr)
2771    JUMPHERE(jump);      {
2772  #elif defined COMPILE_PCRE16      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2773    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2774    jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);      OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2775    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);      OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2776    JUMPHERE(jump);      OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2777    /* Skip low surrogate if necessary. */      OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);
2778    OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xfc00);      OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2779    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, 0xd800);      jump2 = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2780    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);      OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2781    OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 1);      JUMPHERE(jump2);
2782    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);      }
2783  #elif defined COMPILE_PCRE32    else
2784    OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);      add_jump(compiler, &common->utfreadtype8, JUMP(SLJIT_FAST_CALL));
   jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  
   OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  
2785    JUMPHERE(jump);    JUMPHERE(jump);
 #endif /* COMPILE_PCRE[8|16|32] */  
2786    return;    return;
2787    }    }
2788  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
2789  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), 0);  
2790  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  #if !defined COMPILE_PCRE8
 #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  
2791  /* The ctypes array contains only 256 values. */  /* The ctypes array contains only 256 values. */
2792  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2793  jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);
2794  #endif  #endif
2795  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2796  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if !defined COMPILE_PCRE8
2797  JUMPHERE(jump);  JUMPHERE(jump);
2798  #endif  #endif
2799    
2800    #if defined SUPPORT_UTF && defined COMPILE_PCRE16
2801    if (common->utf && update_str_ptr)
2802      {
2803      /* Skip low surrogate if necessary. */
2804      OP2(SLJIT_SUB, TMP2, 0, TMP2, 0, SLJIT_IMM, 0xd800);
2805      jump = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0xdc00 - 0xd800 - 1);
2806      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2807      JUMPHERE(jump);
2808      }
2809    #endif /* SUPPORT_UTF && COMPILE_PCRE16 */
2810  }  }
2811    
2812  static void skip_char_back(compiler_common *common)  static void skip_char_back(compiler_common *common)
# Line 2370  if (common->utf) Line 2844  if (common->utf)
2844  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2845  }  }
2846    
2847  static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpiftrue)  static void check_newlinechar(compiler_common *common, int nltype, jump_list **backtracks, BOOL jumpifmatch)
2848  {  {
2849  /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */  /* Character comes in TMP1. Checks if it is a newline. TMP2 may be destroyed. */
2850  DEFINE_COMPILER;  DEFINE_COMPILER;
2851    struct sljit_jump *jump;
2852    
2853  if (nltype == NLTYPE_ANY)  if (nltype == NLTYPE_ANY)
2854    {    {
2855    add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
2856    add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(jumpifmatch ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));
2857    }    }
2858  else if (nltype == NLTYPE_ANYCRLF)  else if (nltype == NLTYPE_ANYCRLF)
2859    {    {
2860    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_CR);    if (jumpifmatch)
2861    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);      {
2862    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_NL);      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR));
2863    OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
2864    add_jump(compiler, backtracks, JUMP(jumpiftrue ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));      }
2865      else
2866        {
2867        jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
2868        add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL));
2869        JUMPHERE(jump);
2870        }
2871    }    }
2872  else  else
2873    {    {
2874    SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256);    SLJIT_ASSERT(nltype == NLTYPE_FIXED && common->newline < 256);
2875    add_jump(compiler, backtracks, CMP(jumpiftrue ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline));    add_jump(compiler, backtracks, CMP(jumpifmatch ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, common->newline));
2876    }    }
2877  }  }
2878    
# Line 2401  else Line 2882  else
2882  static void do_utfreadchar(compiler_common *common)  static void do_utfreadchar(compiler_common *common)
2883  {  {
2884  /* Fast decoding a UTF-8 character. TMP1 contains the first byte  /* Fast decoding a UTF-8 character. TMP1 contains the first byte
2885  of the character (>= 0xc0). Return char value in TMP1, length - 1 in TMP2. */  of the character (>= 0xc0). Return char value in TMP1, length in TMP2. */
2886  DEFINE_COMPILER;  DEFINE_COMPILER;
2887  struct sljit_jump *jump;  struct sljit_jump *jump;
2888    
2889  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);  sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
2890    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2891    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2892    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2893    OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2894    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2895    
2896  /* Searching for the first zero. */  /* Searching for the first zero. */
2897  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x20);  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800);
2898  jump = JUMP(SLJIT_C_NOT_ZERO);  jump = JUMP(SLJIT_C_NOT_ZERO);
2899  /* Two byte sequence. */  /* Two byte sequence. */
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
2900  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2901  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x1f);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2));
2902    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2903    
2904    JUMPHERE(jump);
2905    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2906    OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800);
2907  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2908  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2909  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));  
 sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  
 JUMPHERE(jump);  
2910    
2911  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10);  OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x10000);
2912  jump = JUMP(SLJIT_C_NOT_ZERO);  jump = JUMP(SLJIT_C_NOT_ZERO);
2913  /* Three byte sequence. */  /* Three byte sequence. */
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
 OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x0f);  
 OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 12);  
 OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  
 OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);  
 OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  
 OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));  
2914  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
2915  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3));
 OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  
 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(2));  
2916  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
 JUMPHERE(jump);  
2917    
2918  /* Four byte sequence. */  /* Four byte sequence. */
2919  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  JUMPHERE(jump);
2920  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x07);  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));
2921  OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 18);  OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);
2922    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2923    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));
2924  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
 OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 12);  
2925  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2926  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(2));  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(4));
2927    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2928    }
2929    
2930    static void do_utfreadchar16(compiler_common *common)
2931    {
2932    /* Fast decoding a UTF-8 character. TMP1 contains the first byte
2933    of the character (>= 0xc0). Return value in TMP1. */
2934    DEFINE_COMPILER;
2935    struct sljit_jump *jump;
2936    
2937    sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);
2938    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2939    OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2940    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2941  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
 OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);  
2942  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2943  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(3));  
2944  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(3));  /* Searching for the first zero. */
2945    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x800);
2946    jump = JUMP(SLJIT_C_NOT_ZERO);
2947    /* Two byte sequence. */
2948    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2949    sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2950    
2951    JUMPHERE(jump);
2952    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x400);
2953    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_NOT_ZERO);
2954    /* This code runs only in 8 bit mode. No need to shift the value. */
2955    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
2956    OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));
2957    OP2(SLJIT_XOR, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x800);
2958    OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 6);
2959  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3f);
2960  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);
2961  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(3));  /* Three byte sequence. */
2962    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(2));
2963  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2964  }  }
2965    
# Line 2472  jump = JUMP(SLJIT_C_NOT_ZERO); Line 2979  jump = JUMP(SLJIT_C_NOT_ZERO);
2979  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));
2980  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
2981  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f);  OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x1f);
2982    /* The upper 5 bits are known at this point. */
2983    compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 0x3);
2984  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);  OP2(SLJIT_SHL, TMP2, 0, TMP2, 0, SLJIT_IMM, 6);
2985  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);  OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3f);
2986  OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);  OP2(SLJIT_OR, TMP2, 0, TMP2, 0, TMP1, 0);
 compare = CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, 255);  
2987  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);  OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), common->ctypes);
2988  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
2989    
2990  JUMPHERE(compare);  JUMPHERE(compare);
2991  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2992  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
 JUMPHERE(jump);  
2993    
2994  /* We only have types for characters less than 256. */  /* We only have types for characters less than 256. */
 OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);  
 OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);  
 OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);  
 sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  
 }  
   
 #elif defined COMPILE_PCRE16  
   
 static void do_utfreadchar(compiler_common *common)  
 {  
 /* Fast decoding a UTF-16 character. TMP1 contains the first 16 bit char  
 of the character (>= 0xd800). Return char value in TMP1, length - 1 in TMP2. */  
 DEFINE_COMPILER;  
 struct sljit_jump *jump;  
   
 sljit_emit_fast_enter(compiler, RETURN_ADDR, 0);  
 jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 0xdc00);  
 /* Do nothing, only return. */  
 sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  
   
2995  JUMPHERE(jump);  JUMPHERE(jump);
2996  /* Combine two 16 bit characters. */  OP1(SLJIT_MOV_UB, TMP2, 0, SLJIT_MEM1(TMP2), (sljit_sw)PRIV(utf8_table4) - 0xc0);
2997  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  OP1(SLJIT_MOV, TMP1, 0, SLJIT_IMM, 0);
2998  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP2, 0);
 OP2(SLJIT_AND, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x3ff);  
 OP2(SLJIT_SHL, TMP1, 0, TMP1, 0, SLJIT_IMM, 10);  
 OP2(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_IMM, 0x3ff);  
 OP2(SLJIT_OR, TMP1, 0, TMP1, 0, TMP2, 0);  
 OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, IN_UCHARS(1));  
 OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, 0x10000);  
2999  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);  sljit_emit_fast_return(compiler, RETURN_ADDR, 0);
3000  }  }
3001    
3002  #endif /* COMPILE_PCRE[8|16] */  #endif /* COMPILE_PCRE8 */
3003    
3004  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
3005    
# Line 2593  if (firstline) Line 3074  if (firstline)
3074      mainloop = LABEL();      mainloop = LABEL();
3075      /* Continual stores does not cause data dependency. */      /* Continual stores does not cause data dependency. */
3076      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, STR_PTR, 0);
3077      read_char(common);      read_char_range(common, common->nlmin, common->nlmax, TRUE);
3078      check_newlinechar(common, common->nltype, &newline, TRUE);      check_newlinechar(common, common->nltype, &newline, TRUE);
3079      CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop);      CMPTO(SLJIT_C_LESS, STR_PTR, 0, STR_END, 0, mainloop);
3080      JUMPHERE(end);      JUMPHERE(end);
# Line 2669  if (newlinecheck) Line 3150  if (newlinecheck)
3150  return mainloop;  return mainloop;
3151  }  }
3152    
3153  #define MAX_N_CHARS 3  #define MAX_N_CHARS 16
3154    #define MAX_N_BYTES 8
3155    
3156  static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)  static SLJIT_INLINE void add_prefix_byte(pcre_uint8 byte, pcre_uint8 *bytes)
3157  {  {
3158  DEFINE_COMPILER;  pcre_uint8 len = bytes[0];
3159  struct sljit_label *start;  int i;
 struct sljit_jump *quit;  
 pcre_uint32 chars[MAX_N_CHARS * 2];  
 pcre_uchar *cc = common->start + 1 + LINK_SIZE;  
 int location = 0;  
 pcre_int32 len, c, bit, caseless;  
 int must_stop;  
3160    
3161  /* We do not support alternatives now. */  if (len == 255)
3162  if (*(common->start + GET(common->start, 1)) == OP_ALT)    return;
3163    return FALSE;  
3164    if (len == 0)
3165      {
3166      bytes[0] = 1;
3167      bytes[1] = byte;
3168      return;
3169      }
3170    
3171    for (i = len; i > 0; i--)
3172      if (bytes[i] == byte)
3173        return;
3174    
3175    if (len >= MAX_N_BYTES - 1)
3176      {
3177      bytes[0] = 255;
3178      return;
3179      }
3180    
3181    len++;
3182    bytes[len] = byte;
3183    bytes[0] = len;
3184    }
3185    
3186    static int scan_prefix(compiler_common *common, pcre_uchar *cc, pcre_uint32 *chars, pcre_uint8 *bytes, int max_chars)
3187    {
3188    /* Recursive function, which scans prefix literals. */
3189    BOOL last, any, caseless;
3190    int len, repeat, len_save, consumed = 0;
3191    pcre_uint32 chr, mask;
3192    pcre_uchar *alternative, *cc_save, *oc;
3193    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3194    pcre_uchar othercase[8];
3195    #elif defined SUPPORT_UTF && defined COMPILE_PCRE16
3196    pcre_uchar othercase[2];
3197    #else
3198    pcre_uchar othercase[1];
3199    #endif
3200    
3201    repeat = 1;
3202  while (TRUE)  while (TRUE)
3203    {    {
3204    caseless = 0;    last = TRUE;
3205    must_stop = 1;    any = FALSE;
3206    switch(*cc)    caseless = FALSE;
3207      switch (*cc)
3208      {      {
     case OP_CHAR:  
     must_stop = 0;  
     cc++;  
     break;  
   
3209      case OP_CHARI:      case OP_CHARI:
3210      caseless = 1;      caseless = TRUE;
3211      must_stop = 0;      case OP_CHAR:
3212        last = FALSE;
3213      cc++;      cc++;
3214      break;      break;
3215    
# Line 2718  while (TRUE) Line 3228  while (TRUE)
3228      cc++;      cc++;
3229      continue;      continue;
3230    
3231        case OP_ASSERT:
3232        case OP_ASSERT_NOT:
3233        case OP_ASSERTBACK:
3234        case OP_ASSERTBACK_NOT:
3235        cc = bracketend(cc);
3236        continue;
3237    
3238        case OP_PLUSI:
3239        case OP_MINPLUSI:
3240        case OP_POSPLUSI:
3241        caseless = TRUE;
3242      case OP_PLUS:      case OP_PLUS:
3243      case OP_MINPLUS:      case OP_MINPLUS:
3244      case OP_POSPLUS:      case OP_POSPLUS:
3245      cc++;      cc++;
3246      break;      break;
3247    
3248        case OP_EXACTI:
3249        caseless = TRUE;
3250      case OP_EXACT:      case OP_EXACT:
3251        repeat = GET2(cc, 1);
3252        last = FALSE;
3253      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
3254      break;      break;
3255    
3256      case OP_PLUSI:      case OP_QUERYI:
3257      case OP_MINPLUSI:      case OP_MINQUERYI:
3258      case OP_POSPLUSI:      case OP_POSQUERYI:
3259      caseless = 1;      caseless = TRUE;
3260        case OP_QUERY:
3261        case OP_MINQUERY:
3262        case OP_POSQUERY:
3263        len = 1;
3264      cc++;      cc++;
3265    #ifdef SUPPORT_UTF
3266        if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
3267    #endif
3268        max_chars = scan_prefix(common, cc + len, chars, bytes, max_chars);
3269        if (max_chars == 0)
3270          return consumed;
3271        last = FALSE;
3272      break;      break;
3273    
3274      case OP_EXACTI:      case OP_KET:
3275      caseless = 1;      cc += 1 + LINK_SIZE;
3276        continue;
3277    
3278        case OP_ALT:
3279        cc += GET(cc, 1);
3280        continue;
3281    
3282        case OP_ONCE:
3283        case OP_ONCE_NC:
3284        case OP_BRA:
3285        case OP_BRAPOS:
3286        case OP_CBRA:
3287        case OP_CBRAPOS:
3288        alternative = cc + GET(cc, 1);
3289        while (*alternative == OP_ALT)
3290          {
3291          max_chars = scan_prefix(common, alternative + 1 + LINK_SIZE, chars, bytes, max_chars);
3292          if (max_chars == 0)
3293            return consumed;
3294          alternative += GET(alternative, 1);
3295          }
3296    
3297        if (*cc == OP_CBRA || *cc == OP_CBRAPOS)
3298          cc += IMM2_SIZE;
3299        cc += 1 + LINK_SIZE;
3300        continue;
3301    
3302        case OP_CLASS:
3303    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3304        if (common->utf && !is_char7_bitset((const pcre_uint8 *)(cc + 1), FALSE)) return consumed;
3305    #endif
3306        any = TRUE;
3307        cc += 1 + 32 / sizeof(pcre_uchar);
3308        break;
3309    
3310        case OP_NCLASS:
3311    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3312        if (common->utf) return consumed;
3313    #endif
3314        any = TRUE;
3315        cc += 1 + 32 / sizeof(pcre_uchar);
3316        break;
3317    
3318    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3319        case OP_XCLASS:
3320    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3321        if (common->utf) return consumed;
3322    #endif
3323        any = TRUE;
3324        cc += GET(cc, 1);
3325        break;
3326    #endif
3327    
3328        case OP_DIGIT:
3329    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3330        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_digit, FALSE))
3331          return consumed;
3332    #endif
3333        any = TRUE;
3334        cc++;
3335        break;
3336    
3337        case OP_WHITESPACE:
3338    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3339        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_space, FALSE))
3340          return consumed;
3341    #endif
3342        any = TRUE;
3343        cc++;
3344        break;
3345    
3346        case OP_WORDCHAR:
3347    #if defined SUPPORT_UTF && defined COMPILE_PCRE8
3348        if (common->utf && !is_char7_bitset((const pcre_uint8 *)common->ctypes - cbit_length + cbit_word, FALSE))
3349          return consumed;
3350    #endif
3351        any = TRUE;
3352        cc++;
3353        break;
3354    
3355        case OP_NOT:
3356        case OP_NOTI:
3357        cc++;
3358        /* Fall through. */
3359        case OP_NOT_DIGIT:
3360        case OP_NOT_WHITESPACE:
3361        case OP_NOT_WORDCHAR:
3362        case OP_ANY:
3363        case OP_ALLANY:
3364    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3365        if (common->utf) return consumed;
3366    #endif
3367        any = TRUE;
3368        cc++;
3369        break;
3370    
3371    #ifdef SUPPORT_UCP
3372        case OP_NOTPROP:
3373        case OP_PROP:
3374    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3375        if (common->utf) return consumed;
3376    #endif
3377        any = TRUE;
3378        cc += 1 + 2;
3379        break;
3380    #endif
3381    
3382        case OP_TYPEEXACT:
3383        repeat = GET2(cc, 1);
3384      cc += 1 + IMM2_SIZE;      cc += 1 + IMM2_SIZE;
3385        continue;
3386    
3387        case OP_NOTEXACT:
3388        case OP_NOTEXACTI:
3389    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
3390        if (common->utf) return consumed;
3391    #endif
3392        any = TRUE;
3393        repeat = GET2(cc, 1);
3394        cc += 1 + IMM2_SIZE + 1;
3395      break;      break;
3396    
3397      default:      default:
3398      must_stop = 2;      return consumed;
     break;  
3399      }      }
3400    
3401    if (must_stop == 2)    if (any)
3402        break;      {
3403    #if defined COMPILE_PCRE8
3404        mask = 0xff;
3405    #elif defined COMPILE_PCRE16
3406        mask = 0xffff;
3407    #elif defined COMPILE_PCRE32
3408        mask = 0xffffffff;
3409    #else
3410        SLJIT_ASSERT_STOP();
3411    #endif
3412    
3413        do
3414          {
3415          chars[0] = mask;
3416          chars[1] = mask;
3417          bytes[0] = 255;
3418    
3419          consumed++;
3420          if (--max_chars == 0)
3421            return consumed;
3422          chars += 2;
3423          bytes += MAX_N_BYTES;
3424          }
3425        while (--repeat > 0);
3426    
3427        repeat = 1;
3428        continue;
3429        }
3430    
3431    len = 1;    len = 1;
3432  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
3433    if (common->utf && HAS_EXTRALEN(cc[0])) len += GET_EXTRALEN(cc[0]);    if (common->utf && HAS_EXTRALEN(*cc)) len += GET_EXTRALEN(*cc);
3434  #endif  #endif
3435    
3436    if (caseless && char_has_othercase(common, cc))    if (caseless && char_has_othercase(common, cc))
3437      {      {
3438      caseless = char_get_othercase_bit(common, cc);  #ifdef SUPPORT_UTF
3439      if (caseless == 0)      if (common->utf)
3440        return FALSE;        {
3441  #ifdef COMPILE_PCRE8        GETCHAR(chr, cc);
3442      caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 8));        if ((int)PRIV(ord2utf)(char_othercase(common, chr), othercase) != len)
3443  #else          return consumed;
3444      if ((caseless & 0x100) != 0)        }
       caseless = ((caseless & 0xff) << 16) | (len - (caseless >> 9));  
3445      else      else
       caseless = ((caseless & 0xff) << 8) | (len - (caseless >> 9));  
3446  #endif  #endif
3447          {
3448          chr = *cc;
3449          othercase[0] = TABLE_GET(chr, common->fcc, chr);
3450          }
3451      }      }
3452    else    else
3453      caseless = 0;      caseless = FALSE;
3454    
3455    while (len > 0 && location < MAX_N_CHARS * 2)    len_save = len;
3456      cc_save = cc;
3457      while (TRUE)
3458      {      {
3459      c = *cc;      oc = othercase;
3460      bit = 0;      do
     if (len == (caseless & 0xff))  
3461        {        {
3462        bit = caseless >> 8;        chr = *cc;
3463        c |= bit;  #ifdef COMPILE_PCRE32
3464          if (SLJIT_UNLIKELY(chr == NOTACHAR))
3465            return consumed;
3466    #endif
3467          add_prefix_byte((pcre_uint8)chr, bytes);
3468    
3469          mask = 0;
3470          if (caseless)
3471            {
3472            add_prefix_byte((pcre_uint8)*oc, bytes);
3473            mask = *cc ^ *oc;
3474            chr |= mask;
3475            }
3476    
3477    #ifdef COMPILE_PCRE32
3478          if (chars[0] == NOTACHAR && chars[1] == 0)
3479    #else
3480          if (chars[0] == NOTACHAR)
3481    #endif
3482            {
3483            chars[0] = chr;
3484            chars[1] = mask;
3485            }
3486          else
3487            {
3488            mask |= chars[0] ^ chr;
3489            chr |= mask;
3490            chars[0] = chr;
3491            chars[1] |= mask;
3492            }
3493    
3494          len--;
3495          consumed++;
3496          if (--max_chars == 0)
3497            return consumed;
3498          chars += 2;
3499          bytes += MAX_N_BYTES;
3500          cc++;
3501          oc++;
3502        }        }
3503        while (len > 0);
3504    
3505      chars[location] = c;      if (--repeat == 0)
3506      chars[location + 1] = bit;        break;
3507    
3508      len--;      len = len_save;
3509      location += 2;      cc = cc_save;
3510      cc++;      }
3511    
3512      repeat = 1;
3513      if (last)
3514        return consumed;
3515      }
3516    }
3517    
3518    static SLJIT_INLINE BOOL fast_forward_first_n_chars(compiler_common *common, BOOL firstline)
3519    {
3520    DEFINE_COMPILER;
3521    struct sljit_label *start;
3522    struct sljit_jump *quit;
3523    pcre_uint32 chars[MAX_N_CHARS * 2];
3524    pcre_uint8 bytes[MAX_N_CHARS * MAX_N_BYTES];
3525    pcre_uint8 ones[MAX_N_CHARS];
3526    int offsets[3];
3527    pcre_uint32 mask;
3528    pcre_uint8 *byte_set, *byte_set_end;
3529    int i, max, from;
3530    int range_right = -1, range_len = 3 - 1;
3531    sljit_ub *update_table = NULL;
3532    BOOL in_range;
3533    
3534    /* This is even TRUE, if both are NULL. */
3535    SLJIT_ASSERT(common->read_only_data_ptr == common->read_only_data);
3536    
3537    for (i = 0; i < MAX_N_CHARS; i++)
3538      {
3539      chars[i << 1] = NOTACHAR;
3540      chars[(i << 1) + 1] = 0;
3541      bytes[i * MAX_N_BYTES] = 0;
3542      }
3543    
3544    max = scan_prefix(common, common->start, chars, bytes, MAX_N_CHARS);
3545    
3546    if (max <= 1)
3547      return FALSE;
3548    
3549    for (i = 0; i < max; i++)
3550      {
3551      mask = chars[(i << 1) + 1];
3552      ones[i] = ones_in_half_byte[mask & 0xf];
3553      mask >>= 4;
3554      while (mask != 0)
3555        {
3556        ones[i] += ones_in_half_byte[mask & 0xf];
3557        mask >>= 4;
3558        }
3559      }
3560    
3561    in_range = FALSE;
3562    for (i = 0; i <= max; i++)
3563      {
3564      if (in_range && (i - from) > range_len && (bytes[(i - 1) * MAX_N_BYTES] <= 4))
3565        {
3566        range_len = i - from;
3567        range_right = i - 1;
3568        }
3569    
3570      if (i < max && bytes[i * MAX_N_BYTES] < 255)
3571        {
3572        if (!in_range)
3573          {
3574          in_range = TRUE;
3575          from = i;
3576          }
3577      }      }
3578      else if (in_range)
3579        in_range = FALSE;
3580      }
3581    
3582    if (range_right >= 0)
3583      {
3584      /* Since no data is consumed (see the assert in the beginning
3585      of this function), this space can be reallocated. */
3586      if (common->read_only_data)
3587        SLJIT_FREE(common->read_only_data);
3588    
3589      common->read_only_data_size += 256;
3590      common->read_only_data = (sljit_uw *)SLJIT_MALLOC(common->read_only_data_size);
3591      if (common->read_only_data == NULL)
3592        return TRUE;
3593    
3594      update_table = (sljit_ub *)common->read_only_data;
3595      common->read_only_data_ptr = (sljit_uw *)(update_table + 256);
3596      memset(update_table, IN_UCHARS(range_len), 256);
3597    
3598    if (location >= MAX_N_CHARS * 2 || must_stop != 0)    for (i = 0; i < range_len; i++)
3599        {
3600        byte_set = bytes + ((range_right - i) * MAX_N_BYTES);
3601        SLJIT_ASSERT(byte_set[0] > 0 && byte_set[0] < 255);
3602        byte_set_end = byte_set + byte_set[0];
3603        byte_set++;
3604        while (byte_set <= byte_set_end)
3605          {
3606          if (update_table[*byte_set] > IN_UCHARS(i))
3607            update_table[*byte_set] = IN_UCHARS(i);
3608          byte_set++;
3609          }
3610        }
3611      }
3612    
3613    offsets[0] = -1;
3614    /* Scan forward. */
3615    for (i = 0; i < max; i++)
3616      if (ones[i] <= 2) {
3617        offsets[0] = i;
3618      break;      break;
3619    }    }
3620    
3621  /* At least two characters are required. */  if (offsets[0] < 0 && range_right < 0)
3622  if (location < 2 * 2)    return FALSE;
3623    
3624    if (offsets[0] >= 0)
3625      {
3626      /* Scan backward. */
3627      offsets[1] = -1;
3628      for (i = max - 1; i > offsets[0]; i--)
3629        if (ones[i] <= 2 && i != range_right)
3630          {
3631          offsets[1] = i;
3632          break;
3633          }
3634    
3635      /* This case is handled better by fast_forward_first_char. */
3636      if (offsets[1] == -1 && offsets[0] == 0 && range_right < 0)
3637      return FALSE;      return FALSE;
3638    
3639      offsets[2] = -1;
3640      /* We only search for a middle character if there is no range check. */
3641      if (offsets[1] >= 0 && range_right == -1)
3642        {
3643        /* Scan from middle. */
3644        for (i = (offsets[0] + offsets[1]) / 2 + 1; i < offsets[1]; i++)
3645          if (ones[i] <= 2)
3646            {
3647            offsets[2] = i;
3648            break;
3649            }
3650    
3651        if (offsets[2] == -1)
3652          {
3653          for (i = (offsets[0] + offsets[1]) / 2; i > offsets[0]; i--)
3654            if (ones[i] <= 2)
3655              {
3656              offsets[2] = i;
3657              break;
3658              }
3659          }
3660        }
3661    
3662      SLJIT_ASSERT(offsets[1] == -1 || (offsets[0] < offsets[1]));
3663      SLJIT_ASSERT(offsets[2] == -1 || (offsets[0] < offsets[2] && offsets[1] > offsets[2]));
3664    
3665      chars[0] = chars[offsets[0] << 1];
3666      chars[1] = chars[(offsets[0] << 1) + 1];
3667      if (offsets[2] >= 0)
3668        {
3669        chars[2] = chars[offsets[2] << 1];
3670        chars[3] = chars[(offsets[2] << 1) + 1];
3671        }
3672      if (offsets[1] >= 0)
3673        {
3674        chars[4] = chars[offsets[1] << 1];
3675        chars[5] = chars[(offsets[1] << 1) + 1];
3676        }
3677      }
3678    
3679    max -= 1;
3680  if (firstline)  if (firstline)
3681    {    {
3682    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
3683      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end);
3684    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);    OP1(SLJIT_MOV, TMP3, 0, STR_END, 0);
3685    OP2(SLJIT_SUB, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
3686      quit = CMP(SLJIT_C_LESS_EQUAL, STR_END, 0, TMP1, 0);
3687      OP1(SLJIT_MOV, STR_END, 0, TMP1, 0);
3688      JUMPHERE(quit);
3689    }    }
3690  else  else
3691    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));    OP2(SLJIT_SUB, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
3692    
3693    #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32)
3694    if (range_right >= 0)
3695      OP1(SLJIT_MOV, RETURN_ADDR, 0, SLJIT_IMM, (sljit_sw)update_table);
3696    #endif
3697    
3698  start = LABEL();  start = LABEL();
3699  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  quit = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3700    
3701  OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(0));  SLJIT_ASSERT(range_right >= 0 || offsets[0] >= 0);
3702  OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));  
3703  OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  if (range_right >= 0)
3704  if (chars[1] != 0)    {
3705    OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);  #if defined COMPILE_PCRE8 || (defined SLJIT_LITTLE_ENDIAN && SLJIT_LITTLE_ENDIAN)
3706  CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right));
3707  if (location > 2 * 2)  #else
3708    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(1));    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(range_right + 1) - 1);
3709  if (chars[3] != 0)  #endif
3710    OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[3]);  
3711  CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[2], start);  #if !(defined SLJIT_CONFIG_X86_32 && SLJIT_CONFIG_X86_32)
3712  if (location > 2 * 2)    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM2(RETURN_ADDR, TMP1), 0);
3713    {  #else
3714    if (chars[5] != 0)    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)update_table);
3715      OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[5]);  #endif
3716    CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[4], start);    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, TMP1, 0);
3717      CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, 0, start);
3718      }
3719    
3720    if (offsets[0] >= 0)
3721      {
3722      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[0]));
3723      if (offsets[1] >= 0)
3724        OP1(MOV_UCHAR, TMP2, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[1]));
3725      OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3726    
3727      if (chars[1] != 0)
3728        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[1]);
3729      CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[0], start);
3730      if (offsets[2] >= 0)
3731        OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), IN_UCHARS(offsets[2] - 1));
3732    
3733      if (offsets[1] >= 0)
3734        {
3735        if (chars[5] != 0)
3736          OP2(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_IMM, chars[5]);
3737        CMPTO(SLJIT_C_NOT_EQUAL, TMP2, 0, SLJIT_IMM, chars[4], start);
3738        }
3739    
3740      if (offsets[2] >= 0)
3741        {
3742        if (chars[3] != 0)
3743          OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, chars[3]);
3744        CMPTO(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, chars[2], start);
3745        }
3746      OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
3747    }    }
 OP2(SLJIT_SUB, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));  
3748    
3749  JUMPHERE(quit);  JUMPHERE(quit);
3750    
3751  if (firstline)  if (firstline)
3752      {
3753      if (range_right >= 0)
3754        OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), common->first_line_end);
3755    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3756      if (range_right >= 0)
3757        {
3758        quit = CMP(SLJIT_C_LESS_EQUAL, STR_PTR, 0, TMP1, 0);
3759        OP1(SLJIT_MOV, STR_PTR, 0, TMP1, 0);
3760        JUMPHERE(quit);
3761        }
3762      }
3763  else  else
3764    OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS((location >> 1) - 1));    OP2(SLJIT_ADD, STR_END, 0, STR_END, 0, SLJIT_IMM, IN_UCHARS(max));
3765  return TRUE;  return TRUE;
3766  }  }
3767    
3768  #undef MAX_N_CHARS  #undef MAX_N_CHARS
3769    #undef MAX_N_BYTES
3770    
3771  static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline)  static SLJIT_INLINE void fast_forward_first_char(compiler_common *common, pcre_uchar first_char, BOOL caseless, BOOL firstline)
3772  {  {
# Line 2942  if (common->nltype == NLTYPE_FIXED && co Line 3872  if (common->nltype == NLTYPE_FIXED && co
3872    JUMPHERE(lastchar);    JUMPHERE(lastchar);
3873    
3874    if (firstline)    if (firstline)
3875      OP1(SLJIT_MOV, STR_END, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE0);      OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3876    return;    return;
3877    }    }
3878    
# Line 2952  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_ Line 3882  firstchar = CMP(SLJIT_C_LESS_EQUAL, STR_
3882  skip_char_back(common);  skip_char_back(common);
3883    
3884  loop = LABEL();  loop = LABEL();
3885  read_char(common);  common->ff_newline_shortcut = loop;
3886    
3887    read_char_range(common, common->nlmin, common->nlmax, TRUE);
3888  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);  lastchar = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);
3889  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)  if (common->nltype == NLTYPE_ANY || common->nltype == NLTYPE_ANYCRLF)
3890    foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);    foundcr = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
# Line 2981  if (firstline) Line 3913  if (firstline)
3913    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);    OP1(SLJIT_MOV, STR_END, 0, TMP3, 0);
3914  }  }
3915    
3916  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks);  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks);
3917    
3918  static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, sljit_uw start_bits, BOOL firstline)  static SLJIT_INLINE void fast_forward_start_bits(compiler_common *common, pcre_uint8 *start_bits, BOOL firstline)
3919  {  {
3920  DEFINE_COMPILER;  DEFINE_COMPILER;
3921  struct sljit_label *start;  struct sljit_label *start;
3922  struct sljit_jump *quit;  struct sljit_jump *quit;
3923  struct sljit_jump *found = NULL;  struct sljit_jump *found = NULL;
3924  jump_list *matches = NULL;  jump_list *matches = NULL;
 pcre_uint8 inverted_start_bits[32];  
 int i;  
3925  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3926  struct sljit_jump *jump;  struct sljit_jump *jump;
3927  #endif  #endif
3928    
 for (i = 0; i < 32; ++i)  
   inverted_start_bits[i] = ~(((pcre_uint8*)start_bits)[i]);  
   
3929  if (firstline)  if (firstline)
3930    {    {
3931    SLJIT_ASSERT(common->first_line_end != 0);    SLJIT_ASSERT(common->first_line_end != 0);
# Line 3014  if (common->utf) Line 3941  if (common->utf)
3941    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
3942  #endif  #endif
3943    
3944  if (!check_class_ranges(common, inverted_start_bits, (inverted_start_bits[31] & 0x80) != 0, &matches))  if (!check_class_ranges(common, start_bits, (start_bits[31] & 0x80) != 0, TRUE, &matches))
3945    {    {
3946  #ifndef COMPILE_PCRE8  #ifndef COMPILE_PCRE8
3947    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);    jump = CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, 255);
# Line 3023  if (!check_class_ranges(common, inverted Line 3950  if (!check_class_ranges(common, inverted
3950  #endif  #endif
3951    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
3952    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
3953    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), start_bits);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)start_bits);
3954    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
3955    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
3956    found = JUMP(SLJIT_C_NOT_ZERO);    found = JUMP(SLJIT_C_NOT_ZERO);
# Line 3167  static void check_wordboundary(compiler_ Line 4094  static void check_wordboundary(compiler_
4094  {  {
4095  DEFINE_COMPILER;  DEFINE_COMPILER;
4096  struct sljit_jump *skipread;  struct sljit_jump *skipread;
4097    jump_list *skipread_list = NULL;
4098  #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF  #if !(defined COMPILE_PCRE8) || defined SUPPORT_UTF
4099  struct sljit_jump *jump;  struct sljit_jump *jump;
4100  #endif  #endif
# Line 3224  else Line 4152  else
4152  JUMPHERE(skipread);  JUMPHERE(skipread);
4153    
4154  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);
4155  skipread = check_str_end(common);  check_str_end(common, &skipread_list);
4156  peek_char(common);  peek_char(common, READ_CHAR_MAX);
4157    
4158  /* Testing char type. This is a code duplication. */  /* Testing char type. This is a code duplication. */
4159  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3265  else Line 4193  else
4193      JUMPHERE(jump);      JUMPHERE(jump);
4194  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
4195    }    }
4196  JUMPHERE(skipread);  set_jumps(skipread_list, LABEL());
4197    
4198  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1);  OP2(SLJIT_XOR | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1);
4199  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);  sljit_emit_fast_return(compiler, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS0);
4200  }  }
4201    
4202  /*  static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, BOOL invert, jump_list **backtracks)
   range format:  
   
   ranges[0] = length of the range (max MAX_RANGE_SIZE, -1 means invalid range).  
   ranges[1] = first bit (0 or 1)  
   ranges[2-length] = position of the bit change (when the current bit is not equal to the previous)  
 */  
   
 static BOOL check_ranges(compiler_common *common, int *ranges, jump_list **backtracks, BOOL readch)  
4203  {  {
4204  DEFINE_COMPILER;  DEFINE_COMPILER;
4205  struct sljit_jump *jump;  int ranges[MAX_RANGE_SIZE];
   
 if (ranges[0] < 0)  
   return FALSE;  
   
 switch(ranges[0])  
   {  
   case 1:  
   if (readch)  
     read_char(common);  
   add_jump(compiler, backtracks, CMP(ranges[1] == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));  
   return TRUE;  
   
   case 2:  
   if (readch)  
     read_char(common);  
   OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);  
   add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));  
   return TRUE;  
   
   case 4:  
   if (ranges[2] + 1 == ranges[3] && ranges[4] + 1 == ranges[5])  
     {  
     if (readch)  
       read_char(common);  
     if (ranges[1] != 0)  
       {  
       add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));  
       add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4]));  
       }  
     else  
       {  
       jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]);  
       add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[4]));  
       JUMPHERE(jump);  
       }  
     return TRUE;  
     }  
   if ((ranges[3] - ranges[2]) == (ranges[5] - ranges[4]) && is_powerof2(ranges[4] - ranges[2]))  
     {  
     if (readch)  
       read_char(common);  
     OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4] - ranges[2]);  
     OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[4]);  
     add_jump(compiler, backtracks, CMP(ranges[1] != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[5] - ranges[4]));  
     return TRUE;  
     }  
   return FALSE;  
   
   default:  
   return FALSE;  
   }  
 }  
   
 static void get_ctype_ranges(compiler_common *common, int flag, int *ranges)  
 {  
 int i, bit, length;  
 const pcre_uint8 *ctypes = (const pcre_uint8*)common->ctypes;  
   
 bit = ctypes[0] & flag;  
 ranges[0] = -1;  
 ranges[1] = bit != 0 ? 1 : 0;  
 length = 0;  
   
 for (i = 1; i < 256; i++)  
   if ((ctypes[i] & flag) != bit)  
     {  
     if (length >= MAX_RANGE_SIZE)  
       return;  
     ranges[2 + length] = i;  
     length++;  
     bit ^= flag;  
     }  
   
 if (bit != 0)  
   {  
   if (length >= MAX_RANGE_SIZE)  
     return;  
   ranges[2 + length] = 256;  
   length++;  
   }  
 ranges[0] = length;  
 }  
   
 static BOOL check_class_ranges(compiler_common *common, const pcre_uint8 *bits, BOOL nclass, jump_list **backtracks)  
 {  
 int ranges[2 + MAX_RANGE_SIZE];  
4206  pcre_uint8 bit, cbit, all;  pcre_uint8 bit, cbit, all;
4207  int i, byte, length = 0;  int i, byte, length = 0;
4208    
4209  bit = bits[0] & 0x1;  bit = bits[0] & 0x1;
4210  ranges[1] = bit;  /* All bits will be zero or one (since bit is zero or one). */
 /* Can be 0 or 255. */  
4211  all = -bit;  all = -bit;
4212    
4213  for (i = 0; i < 256; )  for (i = 0; i < 256; )
# Line 3389  for (i = 0; i < 256; ) Line 4222  for (i = 0; i < 256; )
4222        {        {
4223        if (length >= MAX_RANGE_SIZE)        if (length >= MAX_RANGE_SIZE)
4224          return FALSE;          return FALSE;
4225        ranges[2 + length] = i;        ranges[length] = i;
4226        length++;        length++;
4227        bit = cbit;        bit = cbit;
4228        all = -cbit;        all = -cbit;
# Line 3402  if (((bit == 0) && nclass) || ((bit == 1 Line 4235  if (((bit == 0) && nclass) || ((bit == 1
4235    {    {
4236    if (length >= MAX_RANGE_SIZE)    if (length >= MAX_RANGE_SIZE)
4237      return FALSE;      return FALSE;
4238    ranges[2 + length] = 256;    ranges[length] = 256;
4239    length++;    length++;
4240    }    }
 ranges[0] = length;  
4241    
4242  return check_ranges(common, ranges, backtracks, FALSE);  if (length < 0 || length > 4)
4243      return FALSE;
4244    
4245    bit = bits[0] & 0x1;
4246    if (invert) bit ^= 0x1;
4247    
4248    /* No character is accepted. */
4249    if (length == 0 && bit == 0)
4250      add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4251    
4252    switch(length)
4253      {
4254      case 0:
4255      /* When bit != 0, all characters are accepted. */
4256      return TRUE;
4257    
4258      case 1:
4259      add_jump(compiler, backtracks, CMP(bit == 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4260      return TRUE;
4261    
4262      case 2:
4263      if (ranges[0] + 1 != ranges[1])
4264        {
4265        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4266        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4267        }
4268      else
4269        add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4270      return TRUE;
4271    
4272      case 3:
4273      if (bit != 0)
4274        {
4275        add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
4276        if (ranges[0] + 1 != ranges[1])
4277          {
4278          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4279          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4280          }
4281        else
4282          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4283        return TRUE;
4284        }
4285    
4286      add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[0]));
4287      if (ranges[1] + 1 != ranges[2])
4288        {
4289        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1]);
4290        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
4291        }
4292      else
4293        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1]));
4294      return TRUE;
4295    
4296      case 4:
4297      if ((ranges[1] - ranges[0]) == (ranges[3] - ranges[2])
4298          && (ranges[0] | (ranges[2] - ranges[0])) == ranges[2]
4299          && is_powerof2(ranges[2] - ranges[0]))
4300        {
4301        OP2(SLJIT_OR, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[0]);
4302        if (ranges[2] + 1 != ranges[3])
4303          {
4304          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2]);
4305          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_LESS : SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
4306          }
4307        else
4308          add_jump(compiler, backtracks, CMP(bit != 0 ? SLJIT_C_EQUAL : SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2]));
4309        return TRUE;
4310        }
4311    
4312      if (bit != 0)
4313        {
4314        i = 0;
4315        if (ranges[0] + 1 != ranges[1])
4316          {
4317          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4318          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4319          i = ranges[0];
4320          }
4321        else
4322          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[0]));
4323    
4324        if (ranges[2] + 1 != ranges[3])
4325          {
4326          OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[2] - i);
4327          add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[2]));
4328          }
4329        else
4330          add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[2] - i));
4331        return TRUE;
4332        }
4333    
4334      OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[0]);
4335      add_jump(compiler, backtracks, CMP(SLJIT_C_GREATER_EQUAL, TMP1, 0, SLJIT_IMM, ranges[3] - ranges[0]));
4336      if (ranges[1] + 1 != ranges[2])
4337        {
4338        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]);
4339        add_jump(compiler, backtracks, CMP(SLJIT_C_LESS, TMP1, 0, SLJIT_IMM, ranges[2] - ranges[1]));
4340        }
4341      else
4342        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, ranges[1] - ranges[0]));
4343      return TRUE;
4344    
4345      default:
4346      SLJIT_ASSERT_STOP();
4347      return FALSE;
4348      }
4349  }  }
4350    
4351  static void check_anynewline(compiler_common *common)  static void check_anynewline(compiler_common *common)
# Line 3774  return cc; Line 4712  return cc;
4712  #define SET_TYPE_OFFSET(value) \  #define SET_TYPE_OFFSET(value) \
4713    if ((value) != typeoffset) \    if ((value) != typeoffset) \
4714      { \      { \
4715      if ((value) > typeoffset) \      if ((value) < typeoffset) \
       OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \  
     else \  
4716        OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \        OP2(SLJIT_ADD, typereg, 0, typereg, 0, SLJIT_IMM, typeoffset - (value)); \
4717        else \
4718          OP2(SLJIT_SUB, typereg, 0, typereg, 0, SLJIT_IMM, (value) - typeoffset); \
4719      } \      } \
4720    typeoffset = (value);    typeoffset = (value);
4721    
4722  #define SET_CHAR_OFFSET(value) \  #define SET_CHAR_OFFSET(value) \
4723    if ((value) != charoffset) \    if ((value) != charoffset) \
4724      { \      { \
4725      if ((value) > charoffset) \      if ((value) < charoffset) \
4726        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (value) - charoffset); \        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(charoffset - (value))); \
4727      else \      else \
4728        OP2(SLJIT_ADD, TMP1, 0, TMP1, 0, SLJIT_IMM, charoffset - (value)); \        OP2(SLJIT_SUB, TMP1, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)((value) - charoffset)); \
4729      } \      } \
4730    charoffset = (value);    charoffset = (value);
4731    
# Line 3795  static void compile_xclass_matchingpath( Line 4733  static void compile_xclass_matchingpath(
4733  {  {
4734  DEFINE_COMPILER;  DEFINE_COMPILER;
4735  jump_list *found = NULL;  jump_list *found = NULL;
4736  jump_list **list = (*cc & XCL_NOT) == 0 ? &found : backtracks;  jump_list **list = (cc[0] & XCL_NOT) == 0 ? &found : backtracks;
4737  pcre_int32 c, charoffset;  sljit_uw c, charoffset, max = 256, min = READ_CHAR_MAX;
 const pcre_uint32 *other_cases;  
4738  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
4739  pcre_uchar *ccbegin;  pcre_uchar *ccbegin;
4740  int compares, invertcmp, numberofcmps;  int compares, invertcmp, numberofcmps;
4741    #if defined SUPPORT_UTF && (defined COMPILE_PCRE8 || defined COMPILE_PCRE16)
4742    BOOL utf = common->utf;
4743    #endif
4744    
4745  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4746  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;  BOOL needstype = FALSE, needsscript = FALSE, needschar = FALSE;
4747  BOOL charsaved = FALSE;  BOOL charsaved = FALSE;
4748  int typereg = TMP1, scriptreg = TMP1;  int typereg = TMP1, scriptreg = TMP1;
4749  pcre_int32 typeoffset;  const pcre_uint32 *other_cases;
4750    sljit_uw typeoffset;
4751  #endif  #endif
4752    
4753  /* Although SUPPORT_UTF must be defined, we are  /* Scanning the necessary info. */
4754     not necessary in utf mode even in 8 bit mode. */  cc++;
4755  detect_partial_match(common, backtracks);  ccbegin = cc;
4756  read_char(common);  compares = 0;
4757    if (cc[-1] & XCL_MAP)
 if ((*cc++ & XCL_MAP) != 0)  
4758    {    {
4759    OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);    min = 0;
 #ifndef COMPILE_PCRE8  
   jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);  
 #elif defined SUPPORT_UTF  
   if (common->utf)  
     jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);  
 #endif  
   
   if (!check_class_ranges(common, (const pcre_uint8 *)cc, TRUE, list))  
     {  
     OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);  
     OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);  
     OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);  
     OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);  
     OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);  
     add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));  
     }  
   
 #ifndef COMPILE_PCRE8  
   JUMPHERE(jump);  
 #elif defined SUPPORT_UTF  
   if (common->utf)  
     JUMPHERE(jump);  
 #endif  
   OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);  
 #ifdef SUPPORT_UCP  
   charsaved = TRUE;  
 #endif  
4760    cc += 32 / sizeof(pcre_uchar);    cc += 32 / sizeof(pcre_uchar);
4761    }    }
4762    
 /* Scanning the necessary info. */  
 ccbegin = cc;  
 compares = 0;  
4763  while (*cc != XCL_END)  while (*cc != XCL_END)
4764    {    {
4765    compares++;    compares++;
4766    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4767      {      {
4768      cc += 2;      cc ++;
4769  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4770      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (c > max) max = c;
4771  #endif      if (c < min) min = c;
4772  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4773      needschar = TRUE;      needschar = TRUE;
4774  #endif  #endif
4775      }      }
4776    else if (*cc == XCL_RANGE)    else if (*cc == XCL_RANGE)
4777      {      {
4778      cc += 2;      cc ++;
4779  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4780      if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);      if (c < min) min = c;
4781  #endif      GETCHARINCTEST(c, cc);
4782      cc++;      if (c > max) max = c;
 #ifdef SUPPORT_UTF  
     if (common->utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);  
 #endif  
4783  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4784      needschar = TRUE;      needschar = TRUE;
4785  #endif  #endif
# Line 3881  while (*cc != XCL_END) Line 4789  while (*cc != XCL_END)
4789      {      {
4790      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);      SLJIT_ASSERT(*cc == XCL_PROP || *cc == XCL_NOTPROP);
4791      cc++;      cc++;
4792        if (*cc == PT_CLIST)
4793          {
4794          other_cases = PRIV(ucd_caseless_sets) + cc[1];
4795          while (*other_cases != NOTACHAR)
4796            {
4797            if (*other_cases > max) max = *other_cases;
4798            if (*other_cases < min) min = *other_cases;
4799            other_cases++;
4800            }
4801          }
4802        else
4803          {
4804          max = READ_CHAR_MAX;
4805          min = 0;
4806          }
4807    
4808      switch(*cc)      switch(*cc)
4809        {        {
4810        case PT_ANY:        case PT_ANY:
# Line 3900  while (*cc != XCL_END) Line 4824  while (*cc != XCL_END)
4824        case PT_SPACE:        case PT_SPACE:
4825        case PT_PXSPACE:        case PT_PXSPACE:
4826        case PT_WORD:        case PT_WORD:
4827          case PT_PXGRAPH:
4828          case PT_PXPRINT:
4829          case PT_PXPUNCT:
4830        needstype = TRUE;        needstype = TRUE;
4831        needschar = TRUE;        needschar = TRUE;
4832        break;        break;
4833    
4834        case PT_CLIST:        case PT_CLIST:
4835          case PT_UCNC:
4836        needschar = TRUE;        needschar = TRUE;
4837        break;        break;
4838    
# Line 3917  while (*cc != XCL_END) Line 4845  while (*cc != XCL_END)
4845  #endif  #endif
4846    }    }
4847    
4848    /* We are not necessary in utf mode even in 8 bit mode. */
4849    cc = ccbegin;
4850    detect_partial_match(common, backtracks);
4851    read_char_range(common, min, max, (cc[-1] & XCL_NOT) != 0);
4852    
4853    if ((cc[-1] & XCL_HASPROP) == 0)
4854      {
4855      if ((cc[-1] & XCL_MAP) != 0)
4856        {
4857        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4858        if (!check_class_ranges(common, (const pcre_uint8 *)cc, (((const pcre_uint8 *)cc)[31] & 0x80) != 0, TRUE, &found))
4859          {
4860          OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4861          OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4862          OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4863          OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4864          OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4865          add_jump(compiler, &found, JUMP(SLJIT_C_NOT_ZERO));
4866          }
4867    
4868        add_jump(compiler, backtracks, JUMP(SLJIT_JUMP));
4869        JUMPHERE(jump);
4870    
4871        cc += 32 / sizeof(pcre_uchar);
4872        }
4873      else
4874        {
4875        OP2(SLJIT_SUB, TMP2, 0, TMP1, 0, SLJIT_IMM, min);
4876        add_jump(compiler, (cc[-1] & XCL_NOT) == 0 ? backtracks : &found, CMP(SLJIT_C_GREATER, TMP2, 0, SLJIT_IMM, max - min));
4877        }
4878      }
4879    else if ((cc[-1] & XCL_MAP) != 0)
4880      {
4881      OP1(SLJIT_MOV, TMP3, 0, TMP1, 0);
4882    #ifdef SUPPORT_UCP
4883      charsaved = TRUE;
4884    #endif
4885      if (!check_class_ranges(common, (const pcre_uint8 *)cc, FALSE, TRUE, list))
4886        {
4887    #ifdef COMPILE_PCRE8
4888        SLJIT_ASSERT(common->utf);
4889    #endif
4890        jump = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
4891    
4892        OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
4893        OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
4894        OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
4895        OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
4896        OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
4897        add_jump(compiler, list, JUMP(SLJIT_C_NOT_ZERO));
4898    
4899        JUMPHERE(jump);
4900        }
4901    
4902      OP1(SLJIT_MOV, TMP1, 0, TMP3, 0);
4903      cc += 32 / sizeof(pcre_uchar);
4904      }
4905    
4906  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4907  /* Simple register allocation. TMP1 is preferred if possible. */  /* Simple register allocation. TMP1 is preferred if possible. */
4908  if (needstype || needsscript)  if (needstype || needsscript)
# Line 3958  if (needstype || needsscript) Line 4944  if (needstype || needsscript)
4944  #endif  #endif
4945    
4946  /* Generating code. */  /* Generating code. */
 cc = ccbegin;  
4947  charoffset = 0;  charoffset = 0;
4948  numberofcmps = 0;  numberofcmps = 0;
4949  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3974  while (*cc != XCL_END) Line 4959  while (*cc != XCL_END)
4959    if (*cc == XCL_SINGLE)    if (*cc == XCL_SINGLE)
4960      {      {
4961      cc ++;      cc ++;
4962  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4963    
4964      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4965        {        {
4966        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4967        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_EQUAL);
4968        numberofcmps++;        numberofcmps++;
4969        }        }
4970      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4971        {        {
4972        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4973        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
4974        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
4975        numberofcmps = 0;        numberofcmps = 0;
4976        }        }
4977      else      else
4978        {        {
4979        jump = CMP(SLJIT_C_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset);        jump = CMP(SLJIT_C_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4980        numberofcmps = 0;        numberofcmps = 0;
4981        }        }
4982      }      }
4983    else if (*cc == XCL_RANGE)    else if (*cc == XCL_RANGE)
4984      {      {
4985      cc ++;      cc ++;
4986  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
     if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4987      SET_CHAR_OFFSET(c);      SET_CHAR_OFFSET(c);
4988  #ifdef SUPPORT_UTF      GETCHARINCTEST(c, cc);
4989      if (common->utf)  
       {  
       GETCHARINC(c, cc);  
       }  
     else  
 #endif  
       c = *cc++;  
4990      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))      if (numberofcmps < 3 && (*cc == XCL_SINGLE || *cc == XCL_RANGE))
4991        {        {
4992        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4993        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(numberofcmps == 0 ? SLJIT_MOV : SLJIT_OR, TMP2, 0, numberofcmps == 0 ? SLJIT_UNUSED : TMP2, 0, SLJIT_C_LESS_EQUAL);
4994        numberofcmps++;        numberofcmps++;
4995        }        }
4996      else if (numberofcmps > 0)      else if (numberofcmps > 0)
4997        {        {
4998        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
4999        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5000        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5001        numberofcmps = 0;        numberofcmps = 0;
5002        }        }
5003      else      else
5004        {        {
5005        jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, c - charoffset);        jump = CMP(SLJIT_C_LESS_EQUAL ^ invertcmp, TMP1, 0, SLJIT_IMM, (sljit_sw)(c - charoffset));
5006        numberofcmps = 0;        numberofcmps = 0;
5007        }        }
5008      }      }
# Line 4086  while (*cc != XCL_END) Line 5051  while (*cc != XCL_END)
5051    
5052        case PT_SPACE:        case PT_SPACE:
5053        case PT_PXSPACE:        case PT_PXSPACE:
       if (*cc == PT_SPACE)  
         {  
         OP1(SLJIT_MOV, TMP2, 0, SLJIT_IMM, 0);  
         jump = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, 11 - charoffset);  
         }  
5054        SET_CHAR_OFFSET(9);        SET_CHAR_OFFSET(9);
5055        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 13 - 9);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xd - 0x9);
5056        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5057        if (*cc == PT_SPACE)  
5058          JUMPHERE(jump);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x85 - 0x9);
5059          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5060    
5061          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x9);
5062          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5063    
5064        SET_TYPE_OFFSET(ucp_Zl);        SET_TYPE_OFFSET(ucp_Zl);
5065        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl);        OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Zl);
# Line 4104  while (*cc != XCL_END) Line 5068  while (*cc != XCL_END)
5068        break;        break;
5069    
5070        case PT_WORD:        case PT_WORD:
5071        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, CHAR_UNDERSCORE - charoffset);        OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_UNDERSCORE - charoffset));
5072        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);        OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5073        /* Fall through. */        /* Fall through. */
5074    
# Line 4152  while (*cc != XCL_END) Line 5116  while (*cc != XCL_END)
5116          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP2, 0, SLJIT_IMM, other_cases[2]);
5117          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5118    
5119          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, other_cases[0] - charoffset);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(other_cases[0] - charoffset));
5120          OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_OR | ((other_cases[3] == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5121    
5122          other_cases += 3;          other_cases += 3;
5123          }          }
5124        else        else
5125          {          {
5126          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset));
5127          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5128          }          }
5129    
5130        while (*other_cases != NOTACHAR)        while (*other_cases != NOTACHAR)
5131          {          {
5132          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, *other_cases++ - charoffset);          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(*other_cases++ - charoffset));
5133          OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);          OP_FLAGS(SLJIT_OR | ((*other_cases == NOTACHAR) ? SLJIT_SET_E : 0), TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5134          }          }
5135        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);        jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5136        break;        break;
5137    
5138          case PT_UCNC:
5139          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_DOLLAR_SIGN - charoffset));
5140          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);
5141          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_COMMERCIAL_AT - charoffset));
5142          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5143          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(CHAR_GRAVE_ACCENT - charoffset));
5144          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5145    
5146          SET_CHAR_OFFSET(0xa0);
5147          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, (sljit_sw)(0xd7ff - charoffset));
5148          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5149          SET_CHAR_OFFSET(0);
5150          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xe000 - 0);
5151          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_GREATER_EQUAL);
5152          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5153          break;
5154    
5155          case PT_PXGRAPH:
5156          /* C and Z groups are the farthest two groups. */
5157          SET_TYPE_OFFSET(ucp_Ll);
5158          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
5159          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
5160    
5161          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
5162    
5163          /* In case of ucp_Cf, we overwrite the result. */
5164          SET_CHAR_OFFSET(0x2066);
5165          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
5166          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5167    
5168          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
5169          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5170    
5171          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x180e - 0x2066);
5172          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5173    
5174          JUMPHERE(jump);
5175          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
5176          break;
5177    
5178          case PT_PXPRINT:
5179          /* C and Z groups are the farthest two groups. */
5180          SET_TYPE_OFFSET(ucp_Ll);
5181          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Ll);
5182          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_GREATER);
5183    
5184          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Zs - ucp_Ll);
5185          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_NOT_EQUAL);
5186    
5187          jump = CMP(SLJIT_C_NOT_EQUAL, typereg, 0, SLJIT_IMM, ucp_Cf - ucp_Ll);
5188    
5189          /* In case of ucp_Cf, we overwrite the result. */
5190          SET_CHAR_OFFSET(0x2066);
5191          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x2069 - 0x2066);
5192          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5193    
5194          OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0x061c - 0x2066);
5195          OP_FLAGS(SLJIT_OR, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);
5196    
5197          JUMPHERE(jump);
5198          jump = CMP(SLJIT_C_ZERO ^ invertcmp, TMP2, 0, SLJIT_IMM, 0);
5199          break;
5200    
5201          case PT_PXPUNCT:
5202          SET_TYPE_OFFSET(ucp_Sc);
5203          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_So - ucp_Sc);
5204          OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_LESS_EQUAL);
5205    
5206          SET_CHAR_OFFSET(0);
5207          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, 0xff);
5208          OP_FLAGS(SLJIT_AND, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5209    
5210          SET_TYPE_OFFSET(ucp_Pc);
5211          OP2(SLJIT_SUB | SLJIT_SET_U, SLJIT_UNUSED, 0, typereg, 0, SLJIT_IMM, ucp_Ps - ucp_Pc);
5212          OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_LESS_EQUAL);
5213          jump = JUMP(SLJIT_C_NOT_ZERO ^ invertcmp);
5214          break;
5215        }        }
5216      cc += 2;      cc += 2;
5217      }      }
# Line 4195  int length; Line 5237  int length;
5237  unsigned int c, oc, bit;  unsigned int c, oc, bit;
5238  compare_context context;  compare_context context;
5239  struct sljit_jump *jump[4];  struct sljit_jump *jump[4];
5240    jump_list *end_list;
5241  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
5242  struct sljit_label *label;  struct sljit_label *label;
5243  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5244  pcre_uchar propdata[5];  pcre_uchar propdata[5];
5245  #endif  #endif
5246  #endif  #endif /* SUPPORT_UTF */
5247    
5248  switch(type)  switch(type)
5249    {    {
# Line 4225  switch(type) Line 5268  switch(type)
5268    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
5269    case OP_DIGIT:    case OP_DIGIT:
5270    /* Digits are usually 0-9, so it is worth to optimize them. */    /* Digits are usually 0-9, so it is worth to optimize them. */
   if (common->digits[0] == -2)  
     get_ctype_ranges(common, ctype_digit, common->digits);  
5271    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5272    /* Flip the starting bit in the negative case. */  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5273    if (type == OP_NOT_DIGIT)    if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_digit, FALSE))
5274      common->digits[1] ^= 1;      read_char7_type(common, type == OP_NOT_DIGIT);
5275    if (!check_ranges(common, common->digits, backtracks, TRUE))    else
5276      {  #endif
5277      read_char8_type(common);      read_char8_type(common, type == OP_NOT_DIGIT);
5278      OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit);      /* Flip the starting bit in the negative case. */
5279      add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_digit);
5280      }    add_jump(compiler, backtracks, JUMP(type == OP_DIGIT ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
   if (type == OP_NOT_DIGIT)  
     common->digits[1] ^= 1;  
5281    return cc;    return cc;
5282    
5283    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
5284    case OP_WHITESPACE:    case OP_WHITESPACE:
5285    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5286    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5287      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_space, FALSE))
5288        read_char7_type(common, type == OP_NOT_WHITESPACE);
5289      else
5290    #endif
5291        read_char8_type(common, type == OP_NOT_WHITESPACE);
5292    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_space);
5293    add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_WHITESPACE ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
5294    return cc;    return cc;
# Line 4252  switch(type) Line 5296  switch(type)
5296    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
5297    case OP_WORDCHAR:    case OP_WORDCHAR:
5298    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5299    read_char8_type(common);  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5300      if (common->utf && is_char7_bitset((const pcre_uint8*)common->ctypes - cbit_length + cbit_word, FALSE))
5301        read_char7_type(common, type == OP_NOT_WORDCHAR);
5302      else
5303    #endif
5304        read_char8_type(common, type == OP_NOT_WORDCHAR);
5305    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, ctype_word);
5306    add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_WORDCHAR ? SLJIT_C_ZERO : SLJIT_C_NOT_ZERO));
5307    return cc;    return cc;
5308    
5309    case OP_ANY:    case OP_ANY:
5310    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5311    read_char(common);    read_char_range(common, common->nlmin, common->nlmax, TRUE);
5312    if (common->nltype == NLTYPE_FIXED && common->newline > 255)    if (common->nltype == NLTYPE_FIXED && common->newline > 255)
5313      {      {
5314      jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);      jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, (common->newline >> 8) & 0xff);
5315        end_list = NULL;
5316      if (common->mode != JIT_PARTIAL_HARD_COMPILE)      if (common->mode != JIT_PARTIAL_HARD_COMPILE)
5317        jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);        add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
5318      else      else
5319        jump[1] = check_str_end(common);        check_str_end(common, &end_list);
5320    
5321      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);      OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
5322      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff));      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, common->newline & 0xff));
5323      if (jump[1] != NULL)      set_jumps(end_list, LABEL());
       JUMPHERE(jump[1]);  
5324      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
5325      }      }
5326    else    else
# Line 4315  switch(type) Line 5364  switch(type)
5364  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
5365    case OP_NOTPROP:    case OP_NOTPROP:
5366    case OP_PROP:    case OP_PROP:
5367    propdata[0] = 0;    propdata[0] = XCL_HASPROP;
5368    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;    propdata[1] = type == OP_NOTPROP ? XCL_NOTPROP : XCL_PROP;
5369    propdata[2] = cc[0];    propdata[2] = cc[0];
5370    propdata[3] = cc[1];    propdata[3] = cc[1];
# Line 4327  switch(type) Line 5376  switch(type)
5376    
5377    case OP_ANYNL:    case OP_ANYNL:
5378    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5379    read_char(common);    read_char_range(common, common->bsr_nlmin, common->bsr_nlmax, FALSE);
5380    jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);    jump[0] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_CR);
5381    /* We don't need to handle soft partial matching case. */    /* We don't need to handle soft partial matching case. */
5382      end_list = NULL;
5383    if (common->mode != JIT_PARTIAL_HARD_COMPILE)    if (common->mode != JIT_PARTIAL_HARD_COMPILE)
5384      jump[1] = CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0);      add_jump(compiler, &end_list, CMP(SLJIT_C_GREATER_EQUAL, STR_PTR, 0, STR_END, 0));
5385    else    else
5386      jump[1] = check_str_end(common);      check_str_end(common, &end_list);
5387    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);    OP1(MOV_UCHAR, TMP1, 0, SLJIT_MEM1(STR_PTR), 0);
5388    jump[2] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL);    jump[1] = CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, CHAR_NL);
5389    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));    OP2(SLJIT_ADD, STR_PTR, 0, STR_PTR, 0, SLJIT_IMM, IN_UCHARS(1));
5390    jump[3] = JUMP(SLJIT_JUMP);    jump[2] = JUMP(SLJIT_JUMP);
5391    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
5392    check_newlinechar(common, common->bsr_nltype, backtracks, FALSE);    check_newlinechar(common, common->bsr_nltype, backtracks, FALSE);
5393      set_jumps(end_list, LABEL());
5394    JUMPHERE(jump[1]);    JUMPHERE(jump[1]);
5395    JUMPHERE(jump[2]);    JUMPHERE(jump[2]);
   JUMPHERE(jump[3]);  
5396    return cc;    return cc;
5397    
5398    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
5399    case OP_HSPACE:    case OP_HSPACE:
5400    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5401    read_char(common);    read_char_range(common, 0x9, 0x3000, type == OP_NOT_HSPACE);
5402    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->hspace, JUMP(SLJIT_FAST_CALL));
5403    add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_NOT_HSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));
5404    return cc;    return cc;
# Line 4356  switch(type) Line 5406  switch(type)
5406    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
5407    case OP_VSPACE:    case OP_VSPACE:
5408    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5409    read_char(common);    read_char_range(common, 0xa, 0x2029, type == OP_NOT_VSPACE);
5410    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));    add_jump(compiler, &common->vspace, JUMP(SLJIT_FAST_CALL));
5411    add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(type == OP_NOT_VSPACE ? SLJIT_C_NOT_ZERO : SLJIT_C_ZERO));
5412    return cc;    return cc;
# Line 4455  switch(type) Line 5505  switch(type)
5505      else      else
5506        {        {
5507        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);        OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), LOCALS1, STR_PTR, 0);
5508        read_char(common);        read_char_range(common, common->nlmin, common->nlmax, TRUE);
5509        add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0));        add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, STR_PTR, 0, STR_END, 0));
5510        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));        add_jump(compiler, &common->anynewline, JUMP(SLJIT_FAST_CALL));
5511        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));        add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
# Line 4503  switch(type) Line 5553  switch(type)
5553    else    else
5554      {      {
5555      skip_char_back(common);      skip_char_back(common);
5556      read_char(common);      read_char_range(common, common->nlmin, common->nlmax, TRUE);
5557      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5558      }      }
5559    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4554  switch(type) Line 5604  switch(type)
5604      }      }
5605    else    else
5606      {      {
5607      peek_char(common);      peek_char(common, common->nlmax);
5608      check_newlinechar(common, common->nltype, backtracks, FALSE);      check_newlinechar(common, common->nltype, backtracks, FALSE);
5609      }      }
5610    JUMPHERE(jump[0]);    JUMPHERE(jump[0]);
# Line 4578  switch(type) Line 5628  switch(type)
5628  #endif  #endif
5629      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);      return byte_sequence_compare(common, type == OP_CHARI, cc, &context, backtracks);
5630      }      }
5631    
5632    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
   read_char(common);  
5633  #ifdef SUPPORT_UTF  #ifdef SUPPORT_UTF
5634    if (common->utf)    if (common->utf)
5635      {      {
# Line 4588  switch(type) Line 5638  switch(type)
5638    else    else
5639  #endif  #endif
5640      c = *cc;      c = *cc;
5641    
5642    if (type == OP_CHAR || !char_has_othercase(common, cc))    if (type == OP_CHAR || !char_has_othercase(common, cc))
5643      {      {
5644        read_char_range(common, c, c, FALSE);
5645      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c));      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c));
5646      return cc + length;      return cc + length;
5647      }      }
5648    oc = char_othercase(common, c);    oc = char_othercase(common, c);
5649      read_char_range(common, c < oc ? c : oc, c > oc ? c : oc, FALSE);
5650    bit = c ^ oc;    bit = c ^ oc;
5651    if (is_powerof2(bit))    if (is_powerof2(bit))
5652      {      {
# Line 4601  switch(type) Line 5654  switch(type)
5654      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));      add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, c | bit));
5655      return cc + length;      return cc + length;
5656      }      }
5657    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, c);    jump[0] = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c);
5658    OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);    add_jump(compiler, backtracks, CMP(SLJIT_C_NOT_EQUAL, TMP1, 0, SLJIT_IMM, oc));
5659    OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_IMM, oc);    JUMPHERE(jump[0]);
   OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);  
   add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));  
5660    return cc + length;    return cc + length;
5661    
5662    case OP_NOT:    case OP_NOT:
# Line 4640  switch(type) Line 5691  switch(type)
5691  #endif /* COMPILE_PCRE8 */  #endif /* COMPILE_PCRE8 */
5692        {        {
5693        GETCHARLEN(c, cc, length);        GETCHARLEN(c, cc, length);
       read_char(common);  
5694        }        }
5695      }      }
5696    else    else
5697  #endif /* SUPPORT_UTF */  #endif /* SUPPORT_UTF */
     {  
     read_char(common);  
5698      c = *cc;      c = *cc;
     }  
5699    
5700    if (type == OP_NOT || !char_has_othercase(common, cc))    if (type == OP_NOT || !char_has_othercase(common, cc))
5701        {
5702        read_char_range(common, c, c, TRUE);
5703      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c));      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_IMM, c));
5704        }
5705    else    else
5706      {      {
5707      oc = char_othercase(common, c);      oc = char_othercase(common, c);
5708        read_char_range(common, c < oc ? c : oc, c > oc ? c : oc, TRUE);
5709      bit = c ^ oc;      bit = c ^ oc;
5710      if (is_powerof2(bit))      if (is_powerof2(bit))
5711        {        {
# Line 4672  switch(type) Line 5723  switch(type)
5723    case OP_CLASS:    case OP_CLASS:
5724    case OP_NCLASS:    case OP_NCLASS:
5725    detect_partial_match(common, backtracks);    detect_partial_match(common, backtracks);
5726    read_char(common);  
5727    if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, backtracks))  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5728      bit = (common->utf && is_char7_bitset((const pcre_uint8 *)cc, type == OP_NCLASS)) ? 127 : 255;
5729      read_char_range(common, 0, bit, type == OP_NCLASS);
5730    #else
5731      read_char_range(common, 0, 255, type == OP_NCLASS);
5732    #endif
5733    
5734      if (check_class_ranges(common, (const pcre_uint8 *)cc, type == OP_NCLASS, FALSE, backtracks))
5735      return cc + 32 / sizeof(pcre_uchar);      return cc + 32 / sizeof(pcre_uchar);
5736    
5737  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF && defined COMPILE_PCRE8
5738    jump[0] = NULL;    jump[0] = NULL;
 #ifdef COMPILE_PCRE8  
   /* This check only affects 8 bit mode. In other modes, we  
   always need to compare the value with 255. */  
5739    if (common->utf)    if (common->utf)
 #endif /* COMPILE_PCRE8 */  
5740      {      {
5741      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, bit);
5742      if (type == OP_CLASS)      if (type == OP_CLASS)
5743        {        {
5744        add_jump(compiler, backtracks, jump[0]);        add_jump(compiler, backtracks, jump[0]);
5745        jump[0] = NULL;        jump[0] = NULL;
5746        }        }
5747      }      }
5748  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #elif !defined COMPILE_PCRE8
5749      jump[0] = CMP(SLJIT_C_GREATER, TMP1, 0, SLJIT_IMM, 255);
5750      if (type == OP_CLASS)
5751        {
5752        add_jump(compiler, backtracks, jump[0]);
5753        jump[0] = NULL;
5754        }
5755    #endif /* SUPPORT_UTF && COMPILE_PCRE8 */
5756    
5757    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);    OP2(SLJIT_AND, TMP2, 0, TMP1, 0, SLJIT_IMM, 0x7);
5758    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);    OP2(SLJIT_LSHR, TMP1, 0, TMP1, 0, SLJIT_IMM, 3);
5759    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);    OP1(SLJIT_MOV_UB, TMP1, 0, SLJIT_MEM1(TMP1), (sljit_sw)cc);
5760    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);    OP2(SLJIT_SHL, TMP2, 0, SLJIT_IMM, 1, TMP2, 0);
5761    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);    OP2(SLJIT_AND | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, TMP2, 0);
5762    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));    add_jump(compiler, backtracks, JUMP(SLJIT_C_ZERO));
5763    
5764  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
5765    if (jump[0] != NULL)    if (jump[0] != NULL)
5766      JUMPHERE(jump[0]);      JUMPHERE(jump[0]);
5767  #endif /* SUPPORT_UTF || !COMPILE_PCRE8 */  #endif
5768    
5769    return cc + 32 / sizeof(pcre_uchar);    return cc + 32 / sizeof(pcre_uchar);
5770    
5771  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
# Line 4806  if (context.length > 0) Line 5870  if (context.length > 0)
5870  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);  return compile_char1_matchingpath(common, *cc, cc + 1, backtracks);
5871  }  }
5872    
 static struct sljit_jump *compile_ref_checks(compiler_common *common, pcre_uchar *cc, jump_list **backtracks)  
 {  
 DEFINE_COMPILER;  
 int offset = GET2(cc, 1) << 1;  
   
 OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  
 if (!common->jscript_compat)  
   {  
   if (backtracks == NULL)  
     {  
     /* OVECTOR(1) contains the "string begin - 1" constant. */  
     OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));  
     OP_FLAGS(SLJIT_MOV, TMP2, 0, SLJIT_UNUSED, 0, SLJIT_C_EQUAL);  
     OP2(SLJIT_SUB | SLJIT_SET_E, SLJIT_UNUSED, 0, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));  
     OP_FLAGS(SLJIT_OR | SLJIT_SET_E, TMP2, 0, TMP2, 0, SLJIT_C_EQUAL);  
     return JUMP(SLJIT_C_NOT_ZERO);  
     }  
   add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));  
   }  
 return CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));  
 }  
   
5873  /* Forward definitions. */  /* Forward definitions. */
5874  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);  static void compile_matchingpath(compiler_common *, pcre_uchar *, pcre_uchar *, backtrack_common *);
5875  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);  static void compile_backtrackingpath(compiler_common *, struct backtrack_common *);
# Line 4860  static void compile_backtrackingpath(com Line 5902  static void compile_backtrackingpath(com
5902    
5903  #define BACKTRACK_AS(type) ((type *)backtrack)  #define BACKTRACK_AS(type) ((type *)backtrack)
5904    
5905  static pcre_uchar *compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)  static void compile_dnref_search(compiler_common *common, pcre_uchar *cc, jump_list **backtracks)
5906    {
5907    /* The OVECTOR offset goes to TMP2. */
5908    DEFINE_COMPILER;
5909    int count = GET2(cc, 1 + IMM2_SIZE);
5910    pcre_uchar *slot = common->name_table + GET2(cc, 1) * common->name_entry_size;
5911    unsigned int offset;
5912    jump_list *found = NULL;
5913    
5914    SLJIT_ASSERT(*cc == OP_DNREF || *cc == OP_DNREFI);
5915    
5916    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1));
5917    
5918    count--;
5919    while (count-- > 0)
5920      {
5921      offset = GET2(slot, 0) << 1;
5922      GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5923      add_jump(compiler, &found, CMP(SLJIT_C_NOT_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5924      slot += common->name_entry_size;
5925      }
5926    
5927    offset = GET2(slot, 0) << 1;
5928    GET_LOCAL_BASE(TMP2, 0, OVECTOR(offset));
5929    if (backtracks != NULL && !common->jscript_compat)
5930      add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset), TMP1, 0));
5931    
5932    set_jumps(found, LABEL());
5933    }
5934    
5935    static void compile_ref_matchingpath(compiler_common *common, pcre_uchar *cc, jump_list **backtracks, BOOL withchecks, BOOL emptyfail)
5936  {  {
5937  DEFINE_COMPILER;  DEFINE_COMPILER;
5938  int offset = GET2(cc, 1) << 1;  BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
5939    int offset = 0;
5940  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
5941  struct sljit_jump *partial;  struct sljit_jump *partial;
5942  struct sljit_jump *nopartial;  struct sljit_jump *nopartial;
5943    
5944  OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));  if (ref)
5945  /* OVECTOR(1) contains the "string begin - 1" constant. */    {
5946  if (withchecks && !common->jscript_compat)    offset = GET2(cc, 1) << 1;
5947    add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));    OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
5948      /* OVECTOR(1) contains the "string begin - 1" constant. */
5949      if (withchecks && !common->jscript_compat)
5950        add_jump(compiler, backtracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
5951      }
5952    else
5953      OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
5954    
5955  #if defined SUPPORT_UTF && defined SUPPORT_UCP  #if defined SUPPORT_UTF && defined SUPPORT_UCP
5956  if (common->utf && *cc == OP_REFI)  if (common->utf && *cc == OP_REFI)
5957    {    {
5958    SLJIT_ASSERT(TMP1 == SLJIT_SCRATCH_REG1 && STACK_TOP == SLJIT_SCRATCH_REG2 && TMP2 == SLJIT_SCRATCH_REG3);    SLJIT_ASSERT(TMP1 == SLJIT_SCRATCH_REG1 && STACK_TOP == SLJIT_SCRATCH_REG2 && TMP2 == SLJIT_SCRATCH_REG3);
5959    OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));    if (ref)
5960        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
5961      else
5962        OP1(SLJIT_MOV, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
5963    
5964    if (withchecks)    if (withchecks)
5965      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);      jump = CMP(SLJIT_C_EQUAL, TMP1, 0, TMP2, 0);
5966    
# Line 4902  if (common->utf && *cc == OP_REFI) Line 5985  if (common->utf && *cc == OP_REFI)
5985  else  else
5986  #endif /* SUPPORT_UTF && SUPPORT_UCP */  #endif /* SUPPORT_UTF && SUPPORT_UCP */
5987    {    {
5988    OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);    if (ref)
5989        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1), TMP1, 0);
5990      else
5991        OP2(SLJIT_SUB | SLJIT_SET_E, TMP2, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw), TMP1, 0);
5992    
5993    if (withchecks)    if (withchecks)
5994      jump = JUMP(SLJIT_C_ZERO);      jump = JUMP(SLJIT_C_ZERO);
5995    
# Line 4939  if (jump != NULL) Line 6026  if (jump != NULL)
6026    else    else
6027      JUMPHERE(jump);      JUMPHERE(jump);
6028    }    }
 return cc + 1 + IMM2_SIZE;  
6029  }  }
6030    
6031  static SLJIT_INLINE pcre_uchar *compile_ref_iterator_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent)  static SLJIT_INLINE pcre_uchar *compile_ref_iterator_matchingpath(compiler_common *common, pcre_uchar *cc, backtrack_common *parent)
6032  {  {
6033  DEFINE_COMPILER;  DEFINE_COMPILER;
6034    BOOL ref = (*cc == OP_REF || *cc == OP_REFI);
6035  backtrack_common *backtrack;  backtrack_common *backtrack;
6036  pcre_uchar type;  pcre_uchar type;
6037    int offset = 0;
6038  struct sljit_label *label;  struct sljit_label *label;
6039  struct sljit_jump *zerolength;  struct sljit_jump *zerolength;
6040  struct sljit_jump *jump = NULL;  struct sljit_jump *jump = NULL;
# Line 4956  BOOL minimize; Line 6044  BOOL minimize;
6044    
6045  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);  PUSH_BACKTRACK(sizeof(iterator_backtrack), cc, NULL);
6046    
6047    if (ref)
6048      offset = GET2(cc, 1) << 1;
6049    else
6050      cc += IMM2_SIZE;
6051  type = cc[1 + IMM2_SIZE];  type = cc[1 + IMM2_SIZE];
6052    
6053    SLJIT_COMPILE_ASSERT((OP_CRSTAR & 0x1) == 0, crstar_opcode_must_be_even);
6054  minimize = (type & 0x1) != 0;  minimize = (type & 0x1) != 0;
6055  switch(type)  switch(type)
6056    {    {
# Line 4994  if (!minimize) Line 6088  if (!minimize)
6088    if (min == 0)    if (min == 0)
6089      {      {
6090      allocate_stack(common, 2);      allocate_stack(common, 2);
6091        if (ref)
6092          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6093      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), STR_PTR, 0);
6094      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(1), SLJIT_IMM, 0);
6095      /* Temporary release of STR_PTR. */      /* Temporary release of STR_PTR. */
6096      OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));      OP2(SLJIT_SUB, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
6097      zerolength = compile_ref_checks(common, ccbegin, NULL);      /* Handles both invalid and empty cases. Since the minimum repeat,
6098        is zero the invalid case is basically the same as an empty case. */
6099        if (ref)
6100          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6101        else
6102          {
6103          compile_dnref_search(common, ccbegin, NULL);
6104          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(TMP2), 0);
6105          OP1(SLJIT_MOV, SLJIT_MEM1(SLJIT_LOCALS_REG), POSSESSIVE1, TMP2, 0);
6106          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(TMP2), sizeof(sljit_sw));
6107          }
6108      /* Restore if not zero length. */      /* Restore if not zero length. */
6109      OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));      OP2(SLJIT_ADD, STACK_TOP, 0, STACK_TOP, 0, SLJIT_IMM, sizeof(sljit_sw));
6110      }      }
6111    else    else
6112      {      {
6113      allocate_stack(common, 1);      allocate_stack(common, 1);
6114        if (ref)
6115          OP1(SLJIT_MOV, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset));
6116      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);      OP1(SLJIT_MOV, SLJIT_MEM1(STACK_TOP), STACK(0), SLJIT_IMM, 0);
6117      zerolength = compile_ref_checks(common, ccbegin, &backtrack->topbacktracks);      if (ref)
6118          {
6119          add_jump(compiler, &backtrack->topbacktracks, CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(1)));
6120          zerolength = CMP(SLJIT_C_EQUAL, TMP1, 0, SLJIT_MEM1(SLJIT_LOCALS_REG), OVECTOR(offset + 1));
6121          }
6122        else
6123          {
6124          compile_dnref_search(common, ccbegin, &backtrack->topbacktracks);